Enzymatic synthesis of amoxicillin by penicillin G acylase in the presence of ionic liquids

Article abstract of DOI:10.1039/c2gc36158b

A major obstacle for the industrial implementation of the enzymatic synthesis of β-lactam antibiotics is the limited yield of the product, due to undesirable hydrolytic reactions. This drawback can be partially avoided by reducing the water activity in the medium. Ionic liquids (ILs) have emerged as an alternative to conventional organic media due to their high thermal and chemical stability, negligible vapor pressure, non-flammability, and easy recycling. In this context, this paper assesses the catalytic activity of penicillin G acylase (E.C.3.5.1.11) in the synthesis of amoxicillin using different ILs, all based on the 1-butyl-3-methylimidazolium cation. An increase of 400% in selectivity (synthesis/hydrolysis, S/H ratio) was observed for the reactions carried out with BMI·PF₆ as a cosolvent at 75% (v_IL/v_water) when compared to the totally aqueous medium (using phosphate buffer). This figure reached 350% for BMI·NTf ₂, while for BMI·BF₄ there was only a slight increase in selectivity. The highest conversion of the β-lactam nucleus (6-APA) was achieved using BMI·NTf₂ as a cosolvent at 71% (v/v), more than 36% above the one in water. No deactivation of the enzyme after the reactions was observed in any of the ILs, and the physical integrity of the biocatalyst particles was maintained.

Full text of DOI:10.1039/c2gc36158b

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Green Chemistry
Cite this: Green Chem., 2012, 14, 3146
www.rsc.org/greenchem
PAPER
Enzymatic synthesis of amoxicillin by penicillin G acylase in the presence of
ionic liquids
Sandra Cerqueira Pereira,a Roberta Bussamara,b Graciane Marin,b Raquel Lima Camargo Giordano,a
Jairton Dupontb and Roberto de Campos Giordano*a
Received 25th July 2012, Accepted 6th September 2012
DOI: 10.1039/c2gc36158b
A major obstacle for the industrial implementation of the enzymatic synthesis of β-lactam antibiotics is
the limited yield of the product, due to undesirable hydrolytic reactions. This drawback can be partially
avoided by reducing the water activity in the medium. Ionic liquids (ILs) have emerged as an alternative
to conventional organic media due to their high thermal and chemical stability, negligible vapor pressure,
non-flammability, and easy recycling. In this context, this paper assesses the catalytic activity of penicillin
G acylase (E.C.3.5.1.11) in the synthesis of amoxicillin using different ILs, all based on the 1-butyl-
3-methylimidazolium cation. An increase of 400% in selectivity (synthesis/hydrolysis, S/H ratio) was
observed for the reactions carried out with BMI·PF6 as a cosolvent at 75% (vIL/vwater) when compared to
the totally aqueous medium (using phosphate buffer). This figure reached 350% for BMI·NTf2, while for
BMI·BF4 there was only a slight increase in selectivity. The highest conversion of the β-lactam nucleus
(6-APA) was achieved using BMI·NTf2 as a cosolvent at 71% (v/v), more than 36% above the one in
water. No deactivation of the enzyme after the reactions was observed in any of the ILs, and the physical
integrity of the biocatalyst particles was maintained.
Among the most routinely used pharmaceuticals worldwide,
β-lactam antibiotics occupy a prominent position. Semi-synthetic
Introduction
Penicillin G acylase (PGA, E.C.3.5.1.11), also known as penicillin
acyltransferase, penicillin amidase or penicillin amidohydrolase,
is an important enzyme for the pharmaceutical industry. It is
used for the hydrolysis of penicillin G, producing 6-amino-
penicillanic acid (6-APA), a key molecule in the synthesis of
semi-synthetic penicillins.1 Additionally, PGA can also catalyze
the enzymatic synthesis of semi-synthetic β-lactam antibiotics,
through the condensation of different acyl groups to a β-lactam
nucleus (for example, 6-APA).2
Bacteria, yeasts and fungi produce PGA, but the most used in
industrial processes is the enzyme from Escherichia coli ATC C
11105.3 This enzyme is a heterodimer. The smaller subunit, α,
with 209 amino acids, has a molecular mass of 20 500 Da and
the larger one, β, has 69 000 Da and 557 amino acids.4 The role
of the α subunit is to recognize the side chain of the substrate,
while the terminal serine, essential for the catalytic activity, is in
the β subunit.5 Residues from both subunits are part of the active
site of the enzyme.6
cephalosporins and penicillins such as cephalexin, cefadroxil,
cefazolin, ampicillin, amoxicillin, among many others, represent
approximately 65% of the ever-growing production of anti-
biotics,7 which outpaced 45 000 tons in 2000.8 Amoxicillin
occupies an important place in this market. It has a broad spec-
trum of activity, high solubility, a high rate of absorption and is
stable under acid conditions, allowing the oral administration of
this drug, which resists the gastric pH.9
The current industrial process used to produce these anti-
biotics is a rather drastic chemical route, including the protection
and de-protection of reactive groups. It demands low temp-
eratures (−30 °C), the use of toxic organochloride solvents, and
generates non-recyclable waste.10 Presently there is a tendency
of concentrating the production of β-lactam antibiotics in a few
countries The yields that are achieved by the chemical synthesis
are very high, thus contributing to the low market prices of these
drugs. This is of course an important feature for public health,
especially in non-developed countries. But the tendency towards
stricter environmental regulation opens a window of opportunity
for other, less harmful, technologies.
The enzymatic synthesis of semi-synthetic β-lactam antibiotics
was first proposed by Cole.11 Working under mild conditions of
reaction and generating harmless salts as waste of the purification
process, the enzymatic route became a target of intense research
as a “cleaner” process. Enzymatic reactions are “environmental-
friendly” strategies. Enzymes are biodegradable, and conse-
quently less polluting than chemical catalysts. Nevertheless, the
aDepartment of Chemical Engineering, Federal University of São
Carlos, Via Washington Luiz, km 235, 13.565-905, São Carlos, SP,
Brazil. E-mail: roberto@ufscar.br; Fax: +55 16 3351 8266;
Tel: +55 16 3351 8269
bLaboratory of Molecular Catalysis, Institute of Chemistry, Federal
University of Rio Grande do Sul, Avenue Bento Gonçalves, 9.500,
91.501-970, Porto Alegre, RS, Brazil. E-mail: jairton.dupont@ufrgs.br;
Fax: +55 51 3308 7204; Tel: +55 51 3308 6321
3146 | Green Chem., 2012, 14, 3146–3156
This journal is © The Royal Society of Chemistry 2012

View Article Online
enzymatic synthesis of β-lactam antibiotics is still not eco-
nomically competitive with the fine-tuned conventional chemical
process.2 Indeed, despite the advances in this field, which
allowed some optimism with respect to the economic feasibility
of the enzymatic process, to the best of our knowledge there is
no industry using the enzymatic synthesis of amoxicillin in its
production process over the world.12
To reduce costs of the enzymatic synthesis, two points must
be addressed: increasing the process yield and selectivity; and
reducing the cost of the biocatalyst, through its reuse. Soluble
enzymes usually exhibit lower stability than chemical catalysts
and often cannot be recovered. This fact hinders their application
in industrial practice. This problem can be overcome by enzyme
immobilization techniques, which enhance thermal and opera-
tional stabilities, make the catalyst easy to handle, and prevent
enzyme aggregation. Besides, the feasibility of recovery and
reuse of enzymes immobilized on solid supports reduces opera-
tional costs in industry.
Fig. 1 KCS of amoxicillin catalyzed by PGA. In reactions, PGA acts
as transferase (for synthesis) and as hydrolase, promoting two undesired
side reactions (hydrolysis 1, of the acyl donor derivative, and hydrolysis
2, of the antibiotic).
Semi-synthetic antibiotics can be produced using enzymes via
two approaches: thermodynamic controlled synthesis (TCS),
which is the reverse reaction of the antibiotic hydrolysis; or kine-
tically controlled synthesis (KCS), which requires substrate
activation.13 In the TCS, the yield is constrained by the thermo-
dynamic equilibrium constant of the reaction, while for the KCS
the competition between the different catalytic activities (trans-
ferase, esterase and amidase) of PGA determines the yield and
selectivity of the overall process.
A major problem with the TCS is that the synthesis of amoxi-
cillin is completely shifted towards the hydrolysis of the anti-
biotic in an aqueous medium, at the conditions where the
enzyme is active. Besides, for the enzymatic synthesis of amoxi-
cillin to occur, the amino group of 6-APA and the carboxylic
group of p-hydroxyphenylglycine must be neutral.14 However,
there is no range of pH where both substrates are predominantly
in a non-ionized state, as is requested for the enzyme action.15
In the KCS of β-lactam antibiotics, an activated derivative of
the acyl donor, an ester or amide, reacts with the β-lactam
nucleus (6-APA) and produces the antibiotic, thus circumventing
the need for a neutral carboxylic group in the side chain precur-
sor, as is requested in the TCS. In this reaction, PGA acts as a
transferase. However, this enzyme is also a hydrolase, thus
6-APA and water molecules compete in the nucleophilic attack
to the acyl-enzyme intermediate.2 As a result, selectivity
becomes a problem. Fig. 1 exemplifies the reactions involved in
the KCS of amoxicillin when the p-hydroxyphenylglycine
methyl ester (PHPGME) is used.
selectivity (synthesis/hydrolysis, S/H ratio) towards the antibiotic
is a key question for the economics of the enzymatic process.16
These undesirable side reactions could be partially avoided by
reducing the water activity (aw) in the reaction medium. One
way to reduce the hydrolytic reactions is the use of organic
cosolvents.
In recent years, much effort has been directed towards the
study of the enzymatic synthesis of semi-synthetic penicillins in
the presence of organic cosolvents, either for the TCS or for the
KCS strategy.17–25 However, although some good results have
been reported with respect to yield and selectivity, organic sol-
vents have the disadvantage of being volatile, flammable, toxic,
and harmful to the environment.
In this scenario, ionic liquids (ILs) have emerged as solvents
that may replace the traditional organic media in various
biotechnological processes. One decade ago, to the best of our
knowledge, ILs began to be used as reaction media in
biotransformations.26–28 The hydration reaction of 1,3-cyanobenzene
catalyzed by nitrile hydratase employed 1-butyl-3-methyl-
imidazolium hexafluorophosphate (BMI·PF6).26 Almost simulta-
neously, BMI·PF6 was also used on the synthesis of Z-aspartame,
catalyzed by thermolysin.27 Furthermore, an aqueous solution
of Candida antarctica lipase B (CALB) dissolved in pure
1-ethyl-3-methylimidazolium triflimide (EMI·NTf2) or 1-butyl-
3-methylimidazolium triflimide (BMI·NTf2) was the biocatalyst
for both butyl butyrate synthesis and for the kinetic resolution of
1-phenylethanol by transesterification.28 After these pioneering
works, the use of ILs in enzymatic reactions has been widely
investigated.
ILs are organic salts that exist as liquids below a threshold
temperature (usually around 100 °C).29 ILs have been considered
a promising class of “green solvents” mainly due to their negli-
gible vapor pressure. Moreover, they are non-flammable and
have high thermal and chemical stability.30,31 In addition,
ILs have been recognized as “designer solvents”, since their
physicochemical properties such as hydrophobicity, viscosity,
density, and solubility can be tuned by selecting different combi-
nations of cations and anions, and of the attached substituent as
well.32,33
In these series–parallel reactions, PHPGME reacts with
6-APA to form amoxicillin and methanol, while its hydrolysis
produces p-hydroxyphenylglycine (PHPG) and methanol. The
hydrolysis of amoxicillin generates 6-APA and PHPG. It should
be stressed that p-hydroxyphenylglycine ethyl ester (PHPGEE)
may also be employed, reducing even further the environmental
impact of the process. This substrate was successfully used in
our group (results not shown), but since the methyl ester is the
standard substrate in the literature, it will be used in this work as
well, for the sake of comparison.
Thus, the hydrolytic reactions are the main drawback hinder-
ing the industrial implementation of the enzymatic synthesis
of semi-synthetic penicillins.16 Hence, maximization of the
Typical ILs are based on organic cations paired with a variety
of anions that have a strongly delocalized negative charge.34 The
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3146–3156 | 3147

View Article Online
Table 1 Water activity (aw) for different ionic liquid–water mixtures
(IL–water), measures at 25 °C after being equilibrated for 24 h, with
standard deviations of triplicate measurements
IL–water (v/v) BMI·BF4
BMI·PF6
BMI·NTf2
12.5%
25.0%
37.5%
50.0%
62.5%
75.0%
87.5%
100.0%
0.7950 0.0003 0.9001 0.0006 0.9988 0.0002
0.6487 0.0005 0.8979 0.0011 0.9945 0.0003
0.5765 0.0006 0.8902 0.0002 0.9902 0.0004
0.5321 0.0011 0.8765 0.0014 0.9862 0.0002
0.4703 0.0007 0.8434 0.0003 0.9521 0.0003
0.3987 0.0010 0.7777 0.0008 0.8904 0.0002
0.3467 0.0006 0.6244 0.0012 0.7586 0.0004
0.1705 0.0005 0.3710 0.0009 0.5719 0.0008
Fig. 2 Chemical structure of the ionic liquids commonly used in bio-
technological processes.
imidazolium (IM) based salts are the most investigated ILs in
biotechnology.29 Moreover, bis(trifluoromethylsulfonyl)imide
(NTf2), tetrafluoroborate (BF4), and hexafluorophosphate (PF6)
are by far the anions most often used.35,36 Fig. 2 shows the
chemical structure of these ions.
Table 2 Reichardt’s dye polarity (ENT) values at 25 °C and water
content of the ionic liquids equilibrated at 25 °C in water, both reported
in the literature43,44
In short, enzymatic synthesis of β-lactam antibiotics has not
gained industrial feasibility yet due to the low selectivity of the
reaction, caused by hydrolysis of substrate and product. Conse-
quently, important antibiotics (amoxicillin among them) continue
to be produced by the chemical route, using organochloride
solvents. Thus, using the “greenest solvent”, water, is not an
alternative in practice. With this motivation, this work assessed
the catalytic activity of PGA for the KCS of amoxicillin using
different ILs, all based on dialkylimidazolium cations. The reac-
tion in a totally aqueous medium (using phosphate buffer) was
chosen as a standard for comparison with reactions carried out in
increasing proportions of ILs, in order to evaluate the influence
of these solvents on the performance indexes of the reactions
(6-APA conversion and selectivity).
43
Ionic liquid
EN
Water equilibrated44
T
BMI·BF4
BMI·PF6
BMI·NTf2
0.673
0.667
0.642
Miscible
11 700 ppm
3280 ppm
reducing the volumetric fraction of IL implies higher values of
aw. It is known that aw depends on the specific interactions
between the solvent and water, i.e., aw is higher for a solvent that
contains a more hydrophobic group.40,41 In short, these inter-
actions are favored in hydrophilic solvents, resulting in lower
values of aw.
The miscibility of ILs with water varies widely and unpredict-
ably, since BMI·BF4 is water-miscible, but BMI·PF6 and
BMI·NTf2, which have similar polarity as BMI·BF4, are only
partially miscible in water.42 It was observed that solvents with
similar polarities may have very different values of miscibility in
water. Table 2 presents Reichardt’s dye polarity (ENT) values
obtained at 25 °C and water content of the ILs equilibrated at
25 °C in water, both reported in the literature.43,44
Results and discussion
The behavior of enzymes in non-aqueous solvents is still not
completely understood. Ever since the pioneering studies of
enzymatic catalysis in organic solvents, it is questioned if a
totally aqueous microenvironment is essential for maintaining
the active conformation of enzymes.37 Water molecules are dis-
tributed among the vicinities of the enzyme macromolecule, the
solvent itself (since there is no total immiscibility), and the
support for immobilization (when used). Thus, one important
aspect, more than the bulk concentration of water in the medium,
is its availability to the enzyme molecule, in order to sustain the
catalytic properties.
The amount of water available for the enzyme in the reaction
medium (enzyme hydration), particularly for interaction with the
active site, is closely related to the thermodynamic water activity
in the medium (aw).38 Thus, measurements of aw were under-
taken. Table 1 shows the values of aw for the three IL–water
media used in the kinetic assays, for different bulk compositions
of each medium, measured at 25 °C after being equilibrated
for 24 h. It is worth noting that the bulk “compositions” of the
IL–water mixture are here expressed as volumetric fractions.
According to Table 1, the anions have a great influence on the
IL–water interaction. For all volumetric fractions, the lowest
values of aw are found for BMI·BF4, while higher values are for
BMI·NTf2. Values of aw for the ILs without addition of water,
100% (v/v), are corroborated by an earlier study (0.17 and
0.30 for BMI·BF4 and BMI·PF6, respectively).39 As expected,
Although ILs are considered as “nano-structured” with polar
and non-polar regions, their polarity is similar to short-chain
alcohols.45,46 However, it is important to distinguish between the
concepts of hydrophobicity and polarity because the former is
often related to the miscibility with water, i.e., hydrophobicity
may be considered as a narrower concept of polarity.33 These
observations endorse that the identity of the anion exerts great
influence on the interactions between water and IL molecules.
The ENT empirical polarity scale is based on the shift of the
charge-transfer absorption band of a solvatochromic probe in the
presence of a solvent. Changes in the position of the charge-
transfer absorption band within the visible spectrum are due to
hydrogen bonding between the solvent and the phenoxide
oxygen atom present in Reichardt’s dye.43 An EN value is then
T
determined as a function of the position of the charge-transfer
absorption band.43,45,48 EN values for ILs are dependent on the
T
bonding strength of hydrogen between the cation that composes
the IL and the phenoxide group present in Reichardt’s dye.45
High EN values indicate that ILs exhibit strong hydrogen
T
bonding forces, and solvents that demonstrate the ability for
hydrogen bonding present the potential to interfere with the
enzyme structure.47,48
3148 | Green Chem., 2012, 14, 3146–3156
This journal is © The Royal Society of Chemistry 2012

View Article Online
Table 3 Experimental viscosities (η) of the ionic liquids obtained at
25 °C and data reported in the literature at the same temperature for the
dried ionic liquids44
Ionic liquid
η (cP)
η (cP)44
BMI·BF4
BMI·PF6
BMI·NTf2
91
234
44
219
450
69
According to the solvent polarity scale, higher values of EN
T
match higher solvent polarity.43 Solvatochromic parameters
provide a simple method to determine the polarity of a solvent.48
Therefore, a comparison between the data presented in Tables 1
and 2 shows that the order of miscibility with water (hydrophili-
city), BMI·BF4 > BMI·PF6 > BMI·NTf2, follows the polarity of
these ILs, thus explaining the values of aw found in the present
work.
Table 3 presents the experimental data of viscosity of the ILs,
obtained at 25 °C, together with data reported in the literature.44
It is observed that the rank of viscosities found in this study is
BMI·PF6 > BMI·BF4 > BMI·NTf2. In addition, the higher sym-
metry of the inorganic anions (PF6 or BF4) compared to the
organic anion (NTf2) may play an important role, i.e., the geo-
metry and molar mass of the anions have a strong influence on
the viscosity of this class of IL, since BMI combined with either
PF6 or NTf2 produces ILs with significantly different viscosi-
ties.44 These differences in viscosity are one of the reasons
behind the recent development of ILs based in NTf2, which
are relatively less viscous compared to ILs containing other
anions.49
Fig. 3 Amoxicillin synthesis at 25 °C and pH 6.5 with initial bulk con-
centrations: 50 mM PHPGME, 50 mM 6-APA, and 0.2 g of 260 IU g−1
of PGA preparation in a totally aqueous medium (100 mM phosphate
buffer pH 6.5). In the graphics: (□) PHPGME, (▲) 6-APA, (■) PHPG
and (∇) amoxicillin. Error bars: S.D., estimated from triplicates.
the indices chosen to assess the performance of the enzymatic
synthesis of amoxicillin. It should be recalled that the experi-
mental conditions were the same for the reactions carried out in
phosphate buffer without the presence of any IL, specifically,
0% (vIL/vwater), and for the reactions using ILs as cosolvents.
The IL/water volumetric ratios ranged from 12.5 to 87.5%
(vIL/vwater), besides the assay conducted in the presence of an
unmixed IL, namely, 100% (vIL/vwater).
ILs are more viscous fluids than conventional organic sol-
vents.50 Nevertheless, the high viscosity of the ILs may slow
down conformational changes of proteins, allowing enzymes to
maintain their native structures and activity.51 In industrial pro-
duction processes, the majority of enzymes are immobilized-
stabilized to facilitate handling and to improve their operational
stability. These heterogeneous catalyst particles are subject to
internal and external mass transport limitations, which are
strongly influenced by the viscosity of the reaction medium.52
An important aspect that must be observed in research with
innovative solvents, as is the case of employing ILs for the
process of enzymatic synthesis of amoxicillin, is the experi-
mental reproducibility that can be achieved. Therefore, to illus-
trate the reproducibility of the reactions that were performed in
this work, Fig. 3 shows the evolution of the KCS of amoxicillin
catalyzed by PGA immobilized on Sepabeads® in a totally
aqueous medium, while Fig. 4 shows some reactions with ILs as
cosolvents (in different concentrations of each one of the ILs).
These graphics show that the standard deviation (S.D.) from tri-
plicates was small, and in all cases lower than 5%, evidencing
the high reproducibility of the experiments. Besides, the profiles
of substrates and products follow the expected trend for this
system of series–parallel reactions. It is important to note that
mass balance calculations were made for all the reactions, always
with deviations lower than 5%.
Fig. 5 presents a comparison, in terms of selectivity, of all the
experiments that were performed in this study. It is observed that
the reactions carried out with BMI·BF4 as a cosolvent showed
just a slight increase of selectivity compared to the standard re-
action, 0% (v/v), while in the reactions that were conducted with
BMI·PF6 and BMI·NTf2 there was a significant increase of selec-
tivity. The selectivity profiles were similar for all the ILs
employed in these syntheses, differing simply in magnitude, i.e.,
the selectivity increases with the increase of concentration of the
IL until it reaches the maximum value at 75% (v/v) of the IL. At
higher concentrations, the selectivity decreases. As previously
mentioned, hydrolytic reactions are the main drawback in the
industrial implementation of the enzymatic synthesis of semi-
synthetic penicillins.16 Fig. 5 indicates that with the increase of
concentration of the IL, and consequently with the reduction of
aw in the reaction medium, it is possible to reduce these undesir-
able hydrolytic side reactions. In this sense, the results of selec-
tivity for the reactions performed with BMI·PF6 and with
BMI·NTf2 are promising. It should be noticed that reactors
operating in the fed-batch mode would most certainly be the best
operational solution for the industrial process of synthesis: the
antibiotic would precipitate during the operation, and so it would
be retrieved from the hydrolytic action of the enzyme, while the
ester would be fed as it is consumed for the synthesis.53 Conse-
quently, a much higher selectivity can be achieved than the ones
reported here. However, in order to compare the different
systems using the same basis, the assays presented in this study
The stoichiometry of the reactions shown in Fig. 1 provides
two degrees of freedom for the system, so the conversion of
6-APA and the selectivity (synthesis/hydrolysis, S/H ratio) were
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3146–3156 | 3149

View Article Online
Fig. 5 Selectivity (synthesis/hydrolysis, S/H ratio) after 180 min of
reaction. Amoxicillin syntheses at 25 °C and pH 6.5 with initial bulk
concentrations: 50 mM PHPGME, 50 mM 6-APA, and 0.2 g of 260 IU g−1
of PGA preparation.
Fig. 6 Conversion of 6-APA (%) after 180 min of reaction. Amoxicil-
lin syntheses at 25 °C and pH 6.5 with initial bulk concentrations:
50 mM PHPGME, 50 mM 6-APA, and 0.2 g of 260 IU g−1 of PGA
preparation.
of aw were higher for BMI·NTf2. In addition, BMI·NTf2 is more
hydrophobic than BMI·PF6 (see Table 2). This means that reac-
tions performed with BMI·NTf2 as a cosolvent had more water
in the reaction medium available for the enzyme, which results
in a higher rate of hydrolysis, i.e., lower selectivity, as can be
seen in Fig. 5.
On the other hand, BMI·BF4, the more hydrophilic IL, shows
the lowest values for the selectivity. The high amount of water
miscible in this IL, in this case, did not hamper the hydrolyses as
was observed for BMI·PF6, showing once again that the
phenomena involved in the enzyme–IL–substrates (including
water) interactions are very complex and interconnected.46,47
Thus, the selection of the optimal solvent cannot be based only
on its physical–chemical properties: kinetic tests are essential for
this assessment.
Fig. 4 Amoxicillin syntheses at 25 °C and pH 6.5 with initial bulk con-
centrations: 50 mM PHPGME, 50 mM 6-APA, and 0.2 g of 260 IU g−1
of PGA preparation in (a) 25% (v/v) BMI·BF4, (b) 50% (v/v)
BMI·NTf2, and (c) 75% (v/v) BMI·PF6. In the graphics: (□) PHPGME,
(▲) 6-APA, (■) PHPG and (∇) amoxicillin. Error bars: S.D., estimated from
triplicates.
were performed in batch reactors, only to standardize the
procedure.
An interesting aspect to be highlighted is the comparison of
the selectivity obtained in the reactions conducted in BMI·PF6
and in BMI·NTf2 with the measurements of aw for these
systems. The analysis of the data in Table 1 shows that all values
Conversion is another important variable in the evaluation of
the performance of the enzymatic process. Fig. 6 presents a
3150 | Green Chem., 2012, 14, 3146–3156
This journal is © The Royal Society of Chemistry 2012

View Article Online
comparison, in terms of 6-APA conversion (%), of all the experi-
ments that were performed in this work. The graphic shows that
the conversion decreases sharply in the reactions with BMI·BF4
as a cosolvent in comparison to the standard reaction. The reac-
tions in BMI·PF6 also showed a decrease of conversion with the
increasing of concentration of the IL, but much less pronounced.
The assays with BMI·NTf2 showed an increase of conversion
until 62.5% (v/v) of IL, and for higher volumetric fractions there
was a decrease in conversion.
shows the photomicrographs of the catalyst before being used in
the reactions (Fig. 7a and b) and after having been employed in
the synthesis of amoxicillin with IL as a cosolvent (Fig. 7c and
d). It was observed that the integrity of the particles was per-
fectly preserved.
The viscosity of the ILs also interferes with the observed
activity of the enzyme, mostly because immobilized enzymes
were used, i.e., the process is subjected to mass transfer resist-
ances both in the external film and in the pores of the particles.56
It is important to mention that for pure ILs, 100% (v/v), there
were no detectable reactions taking place. In Fig. 6, an “unex-
pected” plateau in the range 12.5–62.5% (v/v) of the more
hydrophobic ILs, BMI·PF6 and BMI·NTf2, is noted. The com-
parison between the conversion profiles and the values of aw in
Table 1, however, indicates that aw values do not change signifi-
cantly in the range 12.5–62.5% (v/v), i.e., the enzyme hydration
in this region was sufficient to maintain the enzymatic activity
almost unaltered for these two ILs in this region of composition.
PGA immobilized on Sepabeads®, the biocatalyst used in the
present study, is a highly stable catalyst, able to withstand
adverse reaction conditions.54 These supports are robust and suit-
able for industrial purposes, with a very high surface density of
epoxide groups, allowing an intense enzyme–support inter-
action.55 Additionally, in this work the enzymatic activity of
PGAwas measured before and after each reaction, and no deacti-
vation was ever observed. Even when no enzymatic activity was
noticed during the kinetic essay, after a careful washing this
activity was fully restored.
Table 3 shows that the rank of viscosities was BMI·PF6 >
BMI·BF4 > BMI·NTf2. This same rank of viscosities was
reported in the literature, however, the viscosities for the dried
ILs were higher than the values found in this study, showing that
the presence of water has an important effect on the viscosity of
the medium.44 Hence, mass transfer resistance may have corro-
borated for a higher conversion in BMI·NTf2 than in BMI·PF6.
On the other hand, for up to 62.5% (v/v) of BMI·NTf2 the con-
version was higher than in a totally aqueous medium. This is an
example of the complexity of the phenomena involved in the
IL–enzyme interaction.
Unlike the other two ILs, there is a systematic decline of
6-APA conversion with increasing volumetric fraction of
BMI·BF4 (see Fig. 6). In this case, rather than the viscosity of
the BMI·BF4 it is the miscibility in water and polarity that
contribute in a decisive way to these results, because these latter
directly affect aw. Certain solvents have a tendency to remove
water from the vicinity of the enzyme, leading to insufficient
hydration and hence to a decrease in enzymatic activity.57,58 The
extent of this phenomenon is greater for hydrophilic solvents,
and for this reason hydrophobic solvents usually have better per-
formance in enzymatic reactions.38
In order to ensure that the particles of PGA immobilized on
Sepabeads® were not structurally affected by the presence of ILs
in the reaction medium, image analyses were performed through
the technique of scanning electron microscopy (SEM). For this
purpose, the biocatalyst particles were equally washed and dried
under vacuum for posterior fixation and coating with gold. Fig. 7
The great majority of enzymatic processes using ILs reported
in the literature involve reactions catalyzed by lipases.59–65 Little
information is found concerning the reactions catalyzed by PGA.
Among them, a study reported that aw should be close to 0.8 to
guarantee the enzymatic activity of PGA in BMI·PF6 and in
BMI·BF4.39 Fig. 8 cross-references the information provided in
Table 1 with Fig. 5 and 6. Fig. 8a shows that there is a moderate
increase of enzymatic activity with BMI·NTf2 for 0.95 < aw
<
1.00. The other two ILs showed monotonic behavior, with the
conversion decreasing systematically with aw, and the three ILs
show a sensible reduction of the enzymatic activity for aw < 0.8.
Fig. 8b, in turn, indicates that selectivity towards the synthesis
goes through a maximum when aw is reduced, for all the three
systems.
In another study, the stability of PGA from E. coli in its native
form in the presence of ILs at 40 °C was evaluated, and it was
reported that the enzyme half-life in these ILs follows the order
BMI·NTf2 > BMI·PF6 > BMI·BF4.66 These results were later
corroborated.67 Besides, it was not observed any activity for
PGA in the presence of various concentrations of BMI·BF4 (v/v)
for the hydrolysis reaction of the penicillin G, while the results
were promising in BMI·PF6, with PGA showing greater stability
than in water at 10 °C.68 Interestingly, in a more recent study,
PGA was unable to catalyze the synthesis of β-lactam antibiotics
in the presence of N-methylimidazole, which is commonly used
as a precursor to some ILs.69
Fig. 7 Photomicrographs of the catalyst (PGA immobilized on Sepa-
beads®): (a) and (b) SEM of the particles before the reaction, (c) and (d)
SEM after the reaction in the presence of an ionic liquid.
The obtained results for selectivity and conversion indicate
that BMI·BF4 does not seem to be a suitable solvent for carrying
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3146–3156 | 3151

View Article Online
Fig. 9 Amoxicillin synthesis at 25 °C and pH 6.5 with initial bulk con-
centrations: 50 mM PHPGME, 50 mM 6-APA, and 0.2 g of 260 IU g−1
of PGA preparation in 71% (v/v) BMI·NTf2. In the graphics: (□)
PHPGME, (▲) 6-APA, (■) PHPG and (∇) amoxicillin.
The reactions carried out in BMI·PF6 showed the highest
values of selectivity, but this increase was at the expense of con-
version. One probable explanation for the reduction of conver-
sion in reactions with BMI·PF6 could be the high viscosity of
this solvent, which may have reduced the apparent rates of reac-
tion. The viscosity of BMI·PF6 could be reduced by increasing
the reaction temperature or by adding some organic solvent that
did not prejudice the stability of PGA, which is beyond the
purpose of this study. In addition, under certain conditions, the
contact of BMI·PF6 with an aqueous phase may result in
hydrolysis of the PF6 anion generating HF.44,70,71
The IL that has shown the best characteristics for use in
the enzymatic synthesis of amoxicillin was BMI·NTf2, which is
the least viscous, and does not tend to remove water from the
enzyme vicinity, since it is the most hydrophobic of them,
without, however, being responsible for any noticeable defor-
mation of the enzyme structure, preserving the integrity of the
active site. Therefore, a study in the range of 62.5 to 75% (v/v)
of BMI·NTf2 was conducted to determine the concentration limit
that would not reduce the conversion of 6-APA and still preserve
the selectivity of the reaction. Fig. 9 shows the evolution of the
KCS of amoxicillin with 71% (vIL/vwater) of BMI·NTf2, which
showed an increase of 36% in conversion compared to the stan-
dard reaction (phosphate buffer), 0% (v/v), and an increase of
more than 300% in selectivity.
An important point that must be evaluated after the synthesis
of a drug is the feasibility of isolation of the target molecule,
amoxicillin, from the reaction mixture. Fig. 10 presents a typical
chromatogram of a sample that was withdrawn during a reaction
of enzymatic synthesis of amoxicillin using BMI·NTf2 as a
cosolvent.
Fig. 8 Conversion of 6-APA (%) in (a) and selectivity (synthesis/
hydrolysis, S/H ratio) in (b), both after 180 min of reaction for different
water activities (aw). Amoxicillin syntheses at 25 °C and pH 6.5 with
initial bulk concentrations: 50 mM PHPGME, 50 mM 6-APA, and 0.2 g
of 260 IU g−1 of PGA preparation. In the graphics: (▲) BMI·BF4, (○)
BMI·PF6 and (□) BMI·NTf2.
out reactions of synthesis of amoxicillin, since there is not a sig-
nificant increase in selectivity, and a very sharp fall in conversion
occurs with respect to the aqueous system. The most promising
solvents were BMI·PF6 and BMI·NTf2, which presented a very
significant increase in selectivity, reaching a value 400% higher
at 75% (v/v) of BMI·PF6 than in a totally aqueous medium,
while BMI·NTf2 showed an increase of more than 350% for this
same volumetric ratio. However, BMI·NTf2 showed better results
in terms of conversion of 6-APA than BMI·PF6 in the range
12.5–62.5% (v/v).
An increase of selectivity reflects the reduction of the undesir-
able hydrolytic reactions. Thus, a smaller amount of by-product
will be formed, i.e., the cost to recover PHPG and to perform
subsequent synthesis of the PHPGME substrate will be lower.
Similarly, a greater conversion of the β-lactam nucleus (6-APA)
represents lower costs to recycle it. At this point, the comparison
between the concentrations 62.5 and 75% (v/v) of BMI·NTf2
shows a decrease of 36% in the conversion and an increase of
30% in the selectivity. The choice between 62.5 or 75% (vIL/
After the enzymatic synthesis of amoxicillin in the pres-
ence of 71% (vIL/vwater) of BMI·NTf2, the result is a mixture
of crystals of amoxicillin and PHPG, as can be observed
in Fig. 11. A small amount of IL is still retained, probably
adhered to the crystal surface (which was not washed yet at this
point).
The purification of the antibiotic was successful, as is shown
in Fig. 12.
v
water) will depend on a cost analysis of this enzymatic process.
3152 | Green Chem., 2012, 14, 3146–3156
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 10 Typical chromatogram of a sample withdrawn during the enzy-
matic synthesis of amoxicillin, including bands of PHPG (2.9 min), 6-
APA (4.5 min), amoxicillin (7.1 min), PHPGME (9.8 min), and
BMI·NTf2 (13–16 min). The mobile phase was composed of 1.4 g of
SDS, 0.6805 g of KH2PO4, 650 mL of ultrahigh-purity water, and
350 mL of acetonitrile, at pH 3.0. Analyses were carried out at 25 °C
and detection at 225 nm as λmax. Column: Phenomenex Gemini C18,
150 × 4.6 mm, 5 μm.
Fig. 12 Typical chromatogram obtained after the purification step
(washing, solubilization and re-crystallization) of the crystals resulting
from the enzymatic synthesis of amoxicillin in the presence of 71%
(v/v) of BMI·NTf2, with a single band of amoxicillin (7.1 min). The
mobile phase was composed of 1.4 g of SDS, 0.6805 g of KH2PO4,
650 mL of ultrahigh-purity water, and 350 mL of acetonitrile, at pH 3.0.
Analyses were carried out at 25 °C and detection at 225 nm as λmax
Column: Phenomenex Gemini C18, 150 × 4.6 mm, 5 μm.
.
Fig. 11 Typical chromatogram of the crystals formed after the enzy-
matic synthesis of amoxicillin in the presence of 71% (v/v) of
BMI·NTf2, including bands of PHPG (2.9 min), amoxicillin (7.1 min),
and BMI·NTf2 (13–16 min). The mobile phase was composed of 1.4 g
of SDS, 0.6805 g of KH2PO4, 650 mL of ultrahigh-purity water, and
350 mL of acetonitrile, at pH 3.0. Analyses were carried out at 25 °C
and detection at 225 nm as λmax. Column: Phenomenex Gemini C18,
150 × 4.6 mm, 5 μm.
Fig. 13 Mass spectrum of amoxicillin synthesized in 71% (v/v) of
BMI·NTf2 and purified by precipitation in its isoelectric point, obtained
using 9 V of cone voltage and 5 eV of collision energy. Spectral data:
MS (ESI+, triple quadrupole) m/z: 366 [M + H]+, 349 [M + H − NH3]+,
208 [M + H − C6H9NO2S]+, 160 [C6H9NO2S + H]+, and 114
[C6H9NO2S − COOH]+.
A mass spectrum after the purification step confirms the iden-
tity of the synthesized amoxicillin (see Fig. 13). The precursor
ion for amoxicillin was the protonated molecular ion [M + H]+ at
366 m/z. One of the main cleavages for amoxicillin produces a
product ion at 349 m/z, due to the loss of ammonia [M + H −
NH3]+. A typical cleavage among β-lactam antibiotics is the
opening of the β-lactam ring, generating a product ion at 160 m/z
[C6H9NO2S + H]+, and a further loss of the carboxylic group
from the thiazolidine ring moiety at 114 m/z [C6H9NO2S −
COOH]+. The product ion at 208 m/z [M + H − C6H9NO2S]+ is
also a result from the cleavage of the antibiotic nucleus.
crystallize in the fed-batch reactor, together with the side-product
PHPG, and then the IL (containing small amounts of dissolved
reactants) would be filtered and recycled for another run of the
reactor, together with un-reacted 6-APA and PHPGME. The
crystals of amoxicillin and PHPG are then washed, for removing
any trace of IL. This stream is then evaporated (providing pure
water as outflow) and the retained IL is recycled to the reactor,
too. The clean amoxicillin crystals would then be dissolved
using NaOH for increasing the pH, the solid PHPG would be
collected and recycled to an esterification reactor, where it would
react with methanol (or even better, ethanol), to provide the
side chain ester substrate. The antibiotic would be finally
A green industrial process using the IL would yield only a
solution of NaCl as a purge stream. The antibiotic would
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3146–3156 | 3153

View Article Online
re-crystallized, using HCl to lower the pH, thus yielding a NaCl
solution as the only residue. It should be noticed that no degra-
dation of the IL was observed after several recycles in lab scale.
method, based on the formation of a Schiff base after the reac-
tion of 6-APA with PDAB.73 The 6-APA produced in the enzy-
matic hydrolysis of penicillin G (Pen G) reacts with PDAB
generating a colored compound that absorbs light in the visible
region and can be monitored in a spectrophotometer at a wave-
length of 415 nm.
Conclusions
The application of ionic liquids (ILs) based on alkyl-substituted
imidazolium for the enzymatic synthesis of amoxicillin was
assessed. The results clearly showed that the nature of the anion
exerts great influence on the interaction of the IL with water
molecules.
The lowest values of water activity (aw) were found for mix-
tures of water and BMI·BF4, while the highest values were
observed for BMI·NTf2, followed by BMI·PF6. This thermodyn-
amic property directly influenced the selectivity of the reactions,
which were significantly higher in the presence of the ILs, reach-
ing increases up to 400%.
A jacketed batch reactor with mechanical stirring was used in
all experiments. The temperature and pH of the reaction medium
were kept constant during the enzymatic hydrolysis assays. The
operational conditions were as follows: a solution of Pen G 5%
(w/v) in 100 mM phosphate buffer pH 8.0, and 37 °C. One inter-
national unit of enzyme (IU) was defined as the amount of
enzyme that hydrolyzes 1 μmol of Pen G per minute at 37 °C
and pH 8.0.
Amoxicillin syntheses
The viscosity of these solvents also affected the conversion of
6-APA. The overall effect was that conversions in the presence
of BMI·PF6 were lower than in a totally aqueous medium (with
100 mM phosphate buffer pH 6.5, the standard medium for com-
parison), but for BMI·NTf2, they were higher than in the stan-
dard medium, approximately 36%.
The biocatalyst used, PGA immobilized on a Sepabeads®
support, did not deactivate during the reaction and its structure
remained unaffected, allowing enzyme reuse and thus reducing
the cost of the synthetic process.
The isolation of the target molecule amoxicillin from the reac-
tion mixture proved to be feasible using a simple purification
procedure, which generates the harmless salt NaCl.
These results have opened up promising possibilities to follow
in trying to make the enzymatic synthesis of semi-synthetic peni-
cillins industrially feasible. However, further research is clearly
required to explore the process implications of using ILs.
Enzymatic reactions for the synthesis of amoxicillin were con-
ducted under kinetic control. In these experiments, the reaction
between PHPGME and 6-APA was catalyzed by PGA immobi-
lized on Sepabeads®. This catalyst presented an apparent enzy-
matic load of 260 IU g−1 of catalyst. A jacketed batch reactor
with constant mechanical stirring was used in all assays. The
temperature of the reaction medium was kept constant during the
enzymatic synthesis by a thermostatic bath, while the pH was
continuously monitored by a pH meter. The experimental con-
ditions were identical both for the reactions carried out in a
totally aqueous medium and for reactions with each ionic liquid
as a cosolvent.
Initial substrate concentrations were 50 mM for both
PHPGME and 6-APA. In addition, the ratio of enzyme to sub-
strate (E/S) was 52 IU mmol−1 of substrate and the total reaction
volume was 20 mL; thus, the amount of catalyst used in all the
experiments was 0.2 g. Every synthesis was conducted at 25 °C.
The standard for comparison was the reaction in a totally
aqueous medium (100 mM phosphate buffer pH 6.5), without
the presence of any ionic liquid. The ILs employed as cosolvents
were BMI·NTf2, BMI·PF6, and BMI·BF4 at (vIL/vwater).
Throughout the course of the reactions, aliquots were withdrawn
and diluted in the mobile phase for HPLC analysis.
Experimental
Materials
The chemicals, amoxicillin, 6-aminopenicillanic acid (6-APA),
p-hydroxyphenylglycine (PHPG), p-dimethylaminobenzalde-
hyde (PDAB), and p-hydroxyphenylglycine methyl ester hydro-
chloride (PHPGME), were from Sigma-Aldrich. Penicillin G
acylase (PGA) from Escherichia coli was covalently immobi-
lized on a Sepabeads® support (Mitsubishi Chemical Corpor-
ation) by the group of Prof. Guisán, from the Department of
Enzymatic Biocatalysis, Institute of Catalysis, CSIC, Madrid,
Spain, and kindly donated for this work.55 All the ionic liquids
(ILs), 1-butyl-3-methylimidazolium hexafluorophosphate (BMI·
PF6), 1-butyl-3-methylimidazolium tetrafluoroborate (BMI·BF4),
and 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)
imide (BMI·NTf2), were prepared as previously described.72 The
other chemicals were of laboratory grade obtained from different
commercial suppliers.
Amoxicillin purification
At the end of the enzymatic syntheses of amoxicillin in the pres-
ence of 71% (vIL/vwater) of BMI·NTf2, the resulting crystals of
amoxicillin and PHPG were removed from the reaction medium
by filtration, and washed with a saturated solution of amoxicillin,
with the purpose of dragging any trace of IL that could be
adhered. Thereafter, the crystals were solubilized at pH 8.5 and
25 °C. After complete solubilization, amoxicillin was re-crystal-
lized in its isoelectric point, 4.9 at 4 °C. All the steps of the
purification were monitored by HPLC analyses.
Amoxicillin characterization
Enzymatic activity
Amoxicillin was characterized using a mass spectrometer
(Waters Xevo TQ, Massachusetts, U.S.). Direct infusion of
Enzymatic activity of PGA immobilized on Sepabeads® was
measured by the p-dimethylaminobenzaldehyde (PDAB)
amoxicillin solubilized in methanol/water (1 : 1), at 2 μg mL−1
,
3154 | Green Chem., 2012, 14, 3146–3156
This journal is © The Royal Society of Chemistry 2012

View Article Online
Scanning electron microscopy
was performed with a syringe pump. The ionization was carried
out by an electrospray source in the positive mode (ESI+).
A triple quadrupole mass analyzer was used. The mass spectrum
(MS) was obtained using 9 V of cone voltage and 5 eV of
collision energy.
Photomicrographs of the catalyst (PGA immobilized on
Sepabeads®) were obtained through the technique of scanning
electron microscopy (SEM) in
S50 microscope equipped with an EBSD detector. All samples
were fixed in aluminium stubs using carbon tape. The coating of
the samples was done with gold in a BAL-TEC SCD 005
Sputter Coater system. Prepared samples were maintained in a
desiccator until the time of analysis.
a FEI Company Inspect
Analytical method
Concentrations were determined by high performance liquid
chromatography (HPLC) in a Shimadzu LC-6AD with an
SPD-10Avp UV-Vis detector and a Phenomenex Gemini C18
column (150 × 4.6 mm, particle size 5 μm). The chromato-
graphic separation was conducted in isocratic elution with the
mobile phase composed of 35% acetonitrile, 4.85 mM sodium
dodecyl sulfate (SDS), 5 mM potassium phosphate monobasic
anhydrous (KH2PO4), and correction to pH 3.0 using phosphoric
acid (H3PO4).
Acknowledgements
The authors would like to thank research-funding agencies
(CAPES, CNPq and FAPESP), and the group of Prof. Guisán of
the Department of Enzymatic Biocatalysis (Institute of Catalysis,
CSIC, Madrid, Spain) for the biocatalyst.
The eluent was filtered through a 0.45 μm membrane and
degassed with ultrasound before use. Analyses were at 25 °C
and detection at 225 nm with an injection volume of 10 μL and
a flow rate of 1 mL min−1. All samples were dissolved in the
mobile phase. All solvents were of HPLC grade. The com-
ponents of the reaction medium were eluted in the following
order: PHPG, 6-APA, amoxicillin, PHPGME, and IL.
Notes and references
1 A. Parmar, H. Kumar, S. S. Marwaha and J. F. Kennedy, Biotechnol.
Adv., 2000, 18, 289.
2 R. C. Giordano, M. P. A. Ribeiro and R. L. C. Giordano, Biotechnol.
Adv., 2006, 24, 27.
3 E. Calleri, C. Temporini, G. Massolini and G. Caccialanza, J. Pharma-
ceut. Biomed. Anal., 2004, 35, 243.
4 W. Böck, R. Schmid, G. Shumacherm, G. Lang and P. Buckel, FEMS
Microbiol. Lett., 1983, 20, 135.
5 W. Brüns, J. Hoppe, H. Tsai, H. J. Brüning, F. Maywald, J. E. Collins and
H. Mayer, J. Mol. Appl. Genet., 1985, 3, 36.
Performance indices for amoxicillin kinetically-controlled
synthesis
6 K. Chandel, L. V. Rao, M. L. Narasu and O. V. Singh, Enzyme Microb.
Technol., 2008, 42, 199.
7 R. P. Elander, Appl. Microbiol. Biotechnol., 2003, 61, 385.
8 A. Bruggink and P. D. Roy, Synthesis of Beta-lactam Antibiotics, Kluwer
Academic Publishers, Netherlands, 2001.
9 G. N. Rolinson and A. M. Geddes, Int. J. Antimicrob. Agents, 2007, 29,
3.
10 S. Ospina, E. Barzana, O. T. Ramirez and A. Lopez-Munguia, Enzyme
Microb. Technol., 1996, 19, 462; Synthesis of β-lactam antibiotics,
ed. A. Bruggink, Kluwer, Dordrecht, 2001.
The performance of enzymatic synthesis of amoxicillin under
the KCS route was assessed using indices of overall selectivity
(eqn (1), S, synthesis/hydrolysis ratio) and 6-APA conversion
(eqn (2), X6-APA) after 180 min of reaction, as follows:
CAmoxicillin ꢀ CAinmitioaxlicillin
S ¼
ð1Þ
CPHPG ꢀ CPinHitPiaGl
11 M. Cole, Biochem. J., 1969, 115, 747.
initial
12 M. A. Wegman, M. H. A. Janssen, F. van Rantwijk and R. A. Sheldon,
Adv. Synth. Catal., 2001, 343, 1; G. Volpato, R. C. Rodrigues and
R. Fernández-Lafuente, Curr. Med. Chem., 2010, 17, 3855;
T. B. Samanta, Indian J. Biotechnol., 2012, 11, 7.
13 M. B. Diender, A. J. J. Straathof, L. A. M. van der Wielen, C. Ras and
J. J. Heijnen, J. Mol. Catal. B: Enzym., 1998, 5, 249; R. Fernández-
Lafuente, C. M. Rosell, B. Piatkowska and J. M. Guisán, Enzyme
Microb. Technol., 1996, 19, 9.
C
ꢀ C6-APA
6-APA
6-APA
X6-APA
¼
ð2Þ
initial
C
where C is the concentration of amoxicillin, or PHPG or 6-APA,
at the beginning and after 180 min of reaction.
14 L. Margolin, V. K. Svedas and I. V. Berezin, Biochim. Biophys. Acta,
1980, 616, 283.
Water activity
15 H. J. Duggleby, S. P. Tolley, C. P. Hill, E. J. Dodson, G. Dodson and
P. C. E. Moody, Nature, 1995, 373, 264.
Water activity (aw) was measured using an AquaLab Series 4
TEV hygrometer from Decagon Devices. Measurements were
carried out in a sealed dew point sensor at 25 °C, until obtaining
constant readings. All samples were previously equilibrated for
24 h. The hygrometer was continuously calibrated with standard
solutions (Decagon Devices).
16 M. E. Gabor, E. J. de Vries and D. B. Janssen, Enzyme Microb. Technol.,
2005, 36, 182; M. P. A. Ribeiro, A. L. O. Ferreira, R. L. C. Giordano and
R. C. Giordano, J. Mol. Catal. B: Enzym., 2005, 33, 81.
17 R. Fernández-Lafuente, C. M. Rosell and J. M. Guisán, Enzyme Microb.
Technol., 1991, 13, 898.
18 G. M. Kim and S. B. Lee, J. Mol. Catal. B: Enzym., 1996, 1, 181.
19 C. M. Rosell, M. Terreni, R. Fernández-Lafuente and J. M. Guisán,
Enzyme Microb. Technol., 1998, 23, 64.
20 O. Abian, C. Mateo, G. Fernández-Lorente, J. M. Palomo, R. Fernández-
Lafuente and J. M. Guisán, Biocatal. Biotransform., 2001, 19, 489.
21 C. Aguirre, M. Toledo, V. Medina and A. Illanes, Proc. Biochem., 2002,
38, 351.
22 A. Illanes, L. Wilson, C. Altamirano, Z. Cabrera, L. Álvarez and
C. Aguirre, Enzyme Microb. Technol., 2007, 40, 195.
23 S. B. Pan, Q. Wu, C. X. Chen and X. F. Lin, J. Mol. Catal. B: Enzym.,
2008, 54, 13.
Viscosity
Measurements of viscosities (η) of all the ionic liquids were
performed in a Brookfield DV-III Programmable rheometer at
25 °C. The rheometer was previously calibrated with standard
oil (Brookfield).
This journal is © The Royal Society of Chemistry 2012
Green Chem., 2012, 14, 3146–3156 | 3155

View Article Online
24 L. L. Du, Q. Wu, C. H. Chen, B. K. Liu and X. F. Lin, J. Mol. Catal. B:
Enzym., 2009, 58, 208.
50 P. Wasserscheid and T. Welton, Ionic Liquids in Synthesis, Wiley-VCH,
Weinheim, 2003.
25 S. M. Bernardino, P. Fernandes and L. P. Fonseca, Biotechnol. Bioeng.,
2010, 107, 753.
26 S. G. Cull, J. D. Holbrey, V. Vargas-Mora, K. R. Seddon and G. J. Lye,
Biotechnol. Bioeng., 2000, 69, 227.
27 M. Erbeldinger, A. J. Mesiano and A. J. Russell, Biotechnol. Prog.,
2000, 16, 1129.
28 P. Lozano, T. de Diego, D. Carrié, M. Vaultier and J. L. Iborra, Chem.
Commun., 2002, 38, 692.
51 F. van Rantwijk and R. A. Sheldon, Chem. Rev., 2007, 107, 2757.
52 U. Kragl, M. Eckstein and N. Kaftzik, Curr. Opin. Biotechnol., 2002, 13,
565.
53 A. L. O. Ferreira, R. L. C. Giordano and R. C. Giordano, Ind. Eng.
Chem. Res., 2007, 46, 7695.
54 C. Mateo, O. Abian, G. Fernández-Lorente, J. Pedroche, R. Fernández-
Lafuente and J. M. Guisán, Biotechnol. Prog., 2002, 18, 629.
55 C. Mateo, V. Grazú, B. C. C. Pessela, T. Montes, J. M. Palomo, R. Torres,
F. López-Gallego, R. Fernández-Lafuente and J. M. Guisán, Biochem.
Soc. Trans., 2007, 35, 1593.
29 G. Quijano, A. Couvert and A. Amrane, Bioresour. Technol., 2010, 101,
8923.
30 J. Dupont, J. Braz. Chem. Soc., 2004, 15, 341.
31 Q. Zhang, S. Zhang and Y. Deng, Green Chem., 2011, 13, 2619.
32 M. Moniruzzaman, N. Kamiya and M. Goto, Org. Biomol. Chem., 2010,
8, 2887.
33 H. Zhao, J. Chem. Technol. Biotechnol., 2010, 85, 891.
34 M. Filice, J. M. Guisán and J. M. Palomo, Green Chem., 2010, 12, 1365.
35 C. Roosen, P. Müller and L. Greiner, Appl. Microbiol. Biotechnol., 2008,
81, 607.
36 M. Suresh kumar and C. K. Lee, J. Mol. Catal. B: Enzym., 2009, 60, 1.
37 A. Zaks and A. M. Klibanov, Science, 1984, 224, 1249.
38 G. Bell, P. J. Halling, B. D. Moore, J. Partridge and D. G. Rees, Trends
Biotechnol., 1995, 13, 468.
39 A. Basso, S. Cantone, P. Linda and C. Ebert, Green Chem., 2005, 7, 671.
40 C. Ebert, L. Gardossi and P. Linda, Tetrahedron Lett., 1996, 37, 9377.
41 C. Ebert, L. Gardossi and P. Linda, J. Mol. Catal. B: Enzym., 1998, 5,
241.
42 F. van Rantwijk, R. M. Lau and R. A. Sheldon, Trends Biotechnol., 2003,
21, 131.
43 C. Reichardt, Green Chem., 2005, 7, 339.
44 J. G. Huddleston, A. E. Visser, W. M. Reichert, H. D. Willauer,
G. A. Broker and E. D. Rogers, Green Chem., 2001, 3, 156.
45 M. J. Muldoon, C. M. Gordon and L. R. Dunkin, J. Chem. Soc., Perkin
Trans., 2001, 2, 433.
56 M. Moniruzzaman, K. Nakashima, N. Kamiya and M. Goto, Biochem.
Eng. J., 2010, 48, 295.
57 A. Zaks and A. M. Klibanov, J. Biol. Chem., 1988, 263, 3194.
58 P. J. Halling, Enzyme Microb. Technol., 1994, 16, 178.
59 R. M. Lau, F. van Rantwijk, K. R. Seddon and R. A. Sheldon, Org. Lett.,
2000, 2, 4189.
60 S. Park and R. J. Kazlauskas, J. Org. Chem., 2001, 66, 8395.
61 S. J. Nara, J. R. Harjani and M. M. Salunkhe, Tetrahedron Lett., 2002,
43, 2979.
62 P. Lozano, T. de Diego, D. Carrié, M. Vaultier and J. L. Iborra, J. Mol.
Catal. B: Enzym., 2003, 21, 9.
63 M. Gamba, A. A. M. Lapis and J. Dupont, Adv. Synth. Catal., 2008, 350,
160.
64 J. Dupont, P. A. Z. Suarez, M. R. Meneghetti and S. M. P. Meneghetti,
Energy Environ. Sci., 2009, 2, 1258.
65 F. J. Hernández-Fernández, A. P. de los Ríos, L. J. Lozano-Blanco and
C. Godínez, J. Chem. Technol. Biotechnol., 2010, 85, 1423.
66 P. de los Ríos, F. J. Hernández-Fernández, M. Rubio, D. Gómez and
G. Víllora, J. Chem. Technol. Biotechnol., 2007, 82, 190.
67 F. J. Hernández-Fernández, A. P. de los Ríos, F. Tomás-Alonso,
D. Gómez and G. Víllora, Can. J. Chem. Eng., 2009, 87, 910.
68 W. Zhang, D. Wei, X. Yang and Q. Song, Bioprocess Biosyst. Eng., 2006,
29, 379.
46 U. Schröder, J. D. Wadhawan, R. G. Compton, F. Marcken,
P. A. Z. Suarez, C. S. Consorti, R. F. de Souza and J. Dupont, New
J. Chem., 2000, 24, 1009.
69 B. Liu, Q. Wu, D. Lv and X. Lin, J. Biotechnol., 2011, 153, 111.
70 J. Dupont, P. A. Z. Suarez, A. P. Umpierre and R. F. de Souza, Catal.
Lett., 2001, 73, 211.
47 J. N. A. C. Lopes and A. A. H. Pádua, J. Phys. Chem. B, 2006, 110, 3330.
48 J. Kaar, A. Jesionowski, J. Berberich, R. Moulton and A. Russell, J. Am.
Chem. Soc., 2003, 125, 4125.
71 P. Peletier, J. Durand and L. Cot, Z. Anorg. Allg. Chem., 1990, 581, 190.
72 C. C. Cassol, G. Ebeling, B. Ferreira and J. Dupont, Adv. Synth. Catal.,
2006, 348, 243.
49 J. Jacquemin, P. Husson, A. A. H. Pádua and V. Majer, Green Chem.,
2006, 8, 172.
73 K. Balasingham, D. Warburton, P. Dunnill and M. D. Lilly, Biochim.
Biophys. Acta, 1972, 276, 250.
3156 | Green Chem., 2012, 14, 3146–3156
This journal is © The Royal Society of Chemistry 2012