Enzymatic hydrolysis of β-lactam antibiotics at low pH in a two-phase "aqueous solution - Water-immiscible organic solvent" system

Article abstract of DOI:10.1139/v02-096

The application of the two-phase "aqueous solution - water-immiscible organic solvent" system is suggested not for effective biocatalytic synthesis, but for hydrolytic purposes. Enzymatic hydrolysis of benzylpenicillin and N-phenylacetamidodesacetoxycepha

Full text of DOI:10.1139/v02-096

699
Enzymatic hydrolysis of -lactam antibiotics
at low pH in a two-phase “ aqueous solution –
water-immiscible organic solvent” system
Ghermes G. Chilov and Vytas K. Švedas
Abstract: The application of the two-phase “aqueous solution – water-immiscible organic solvent” system is suggested
not for effective biocatalytic synthesis, but for hydrolytic purposes. Enzymatic hydrolysis of benzylpenicillin and N-
phenylacetamidodesacetoxycephalosporanic acid to corresponding antibiotic nuclei 6-aminopenicillanic and 7-
aminodesacetoxycephalosporanic acids in a two-phase water–butylacetate system at pH 3–4 is proposed as an alterna-
tive to the biocatalytic hydrolysis in an alkaline medium. An experimental study has been performed and a model has
been developed, which describes the influence of pH, phase volume ratio, thermodynamic constants, and initial antibi-
otic concentration on the effectiveness of their hydrolysis in a two-phase “aqueous solution – water-immiscible organic
solvent” system. The thermodynamic evaluation of penicillin G and 7-phenylacetamidodesacetoxycephalosporanic acid
hydrolysis at low pH in a two-phase aqueous solution – water-immiscible organic solvent system has demonstrated
high practical potential. The suggested approach allows for the exclusion of several technological steps during the
transformation of natural ꢀ-lactam antibiotics to their semi-synthetic analogues: alkaline extraction of the biosynthetic
antibiotic from butylacetate followed by its enzymatic hydrolysis at pH 7.5–8.0 and further acidification of the reaction
mixture, which results in the precipitation of the antibiotic nucleus. Experimental observations also revealed a specific
feature of this process: the kinetic supersaturation of the antibiotic nucleus slows down the attainment of the equilib-
rium, which should be taken into account when further developing this approach.
Key words: enzymatic hydrolysis, ꢀ-lactam antibiotic nuclei, two-phase systems, supersaturation, penicillin acylase.
Résumé : On propose l’utilisation du système à deux phases « solution aqueuse – solvant organique immiscible dans
l’eau », non pas pour des synthèses biocatalytiques efficaces, mais pour des fins d’hydrolyse. On propose une hydro-
lyse enzymatique dans le système à deux phases eau–acétate de butyle à un pH allant de 3 à 4 comme alternative à
l’hydrolyse biocatalytique en milieu alcalin de la benzylpénicilline et de l’acide N-phénylacétamidodésacétoxy-
céphalosporanique conduisant aux noyaux des antibiotiques correspondants, les acides 6-aminopénicillanique et 7-
aminodésacétoxycéphalosporanique. On a effectué une étude expérimentale et on a développé un modèle qui permet de
décrire l’influence du pH, du rapport des volumes dans la phase, des constantes thermodynamiques et de la concentra-
tion initiale d’antibiotique sur l’efficacité de leurs hydrolyses dans un système à deux phases « solution aqueuse –
solvant organique à deux phases ». L’évaluation thermodynamique de l’hydrolyse de la pénicilline G et de l’acide 7-
phénylacétamidodésacétoxycéphalosporanique à faible pH, dans un un système à deux phases « solution aqueuse –
solvant organique à deux phases » a permis de démontrer le caractère très pratique de ce système. L’approche suggérée
permet d’exclure plusieurs étapes technologiques de la transformation des antibiotiques ꢀ-lactames naturels en leurs an-
alogues semi-synthétiques, telle l’extraction alcaline de l’antibiotique biosynthétique de l’acétate de butyle, suivie de
son hydrolyse enzymatique à pH de 7,5 à 8,0 et de l’acidification subséquente du milieu réactionnel pour précipiter le
noyau de l’antibiotique. Une réalisation expérimentale de cette approche a aussi permis de mettre en évidence une
caractéristique spécifique du procédé: dans le développement de cette approche, on doit tenir compte du fait que la
sursaturation cinétique du noyau antibiotique ralentit l’obtention de l’équilibre.
Mots clés : hydrolyse enzymatique, noyau antibiotique ꢀ-lactame, systèmes à deux phases, sursaturation, acylase de
pénicilline.
[Traduit par la Rédaction]
Chilov and Švedas 707
Received 14 January 2002. Published on the NRC Research Press Web site at http://canjchem.nrc.ca on 5 June 2002.
Dedicated to Professor J. Bryan Jones on the occasion of his 65th birthday.
G.G. Chilov and V.K. Švedas.1 Department of Biokinetics, Belozersky Institute of Physicochemical Biology, Lomonosov Moscow
State University, 119899 Moscow, Russia.
1Corresponding author (e-mail: vytas@belozersky.msu.ru).
Can. J. Chem. 80: 699–707 (2002)
DOI: 10.1139/V02-096
© 2002 NRC Canada

700
Can. J. Chem. Vol. 80, 2002
Fig. 1. Penicillin acylase catalyzed hydrolysis of benzylpenicillin (Pen G) and 7-phenylacetamidodesacetoxycephalosporanic acid
(CephG) to antibiotic nuclei (6-APA and 7-ADCA, respectively) and phenylacetic acid (PAA).
Penicillin acylase
+
PAA
6-APA
PenG
Penicillin acylase
+
CephG
PAA
7-ADCA
biocatalytic reactor, as recently proposed by Straathof et al.
(8)), it is necessary to conceive the thermodynamic and
kinetic parameters of the system, which influence the effec-
tiveness of this process. Here, a more detailed analysis of
thermodynamic model of ꢀ-lactam antibiotic hydrolysis in a
two-phase aqueous solution – water-immiscible organic sol-
vent system is described and some kinetic features of the
process are shown to be crucial. Critical evaluation of the
experimental results and mathematic modeling demonstrate
the potential of the proposed process.
Introduction
In a technological scheme of natural penicillin conversion
to its semi-synthetic analogues, penicillin G (Pen G) is first
extracted from the fermentation broth by organic solvent
(butylacetate, amylacetate) at pH 2.0–2.5 (1). After back-
extraction into aqueous solution, it is enzymatically hydro-
lyzed at pH 7–8 (Fig. 1). After acidification of the reaction
mixture, the antibiotic nucleus is isolated by precipitation at
its isoelectric point (pH 3.5) (2). Attractiveness of such pro-
cess stems mainly from thermodynamic considerations,
where antibiotic hydrolysis could reach nearly quantitative
yields above pH 7 (3). Moreover, the optimum pH (7.5–8.0)
of penicillin acylase catalytic activity coincides well with
these conditions (4). The major shortcomings of the above-
mentioned Pen G conversion to 6-aminopenicillanic acid
(6-APA) include numerous pH shifts leading to formation of
waste inorganic salts and degradation of a ꢀ-lactam ring in
an alkaline medium. Additionally, kinetic factors such as
strong product inhibition (competitive by phenylacetic acid
and non-competitive by 6-APA) should be borne in mind
(5). These drawbacks can, however, be avoided in the pro-
cess presented in this work — the hydrolysis of biosynthetic
ꢀ-lactam antibiotics Pen G and 7-phenylacetamidodesace-
toxycephalosporanic acid (Ceph G) at pH 3–4 in a two-
phase water–butylacetate system. Butylacetate containing
the acidic form of the antibiotic can be directly added to an
aqueous phase containing penicillin acylase, which is fol-
lowed by partitioning of antibiotic between two phases, hy-
drolysis of the antibiotic in an aqueous phase, precipitation
of an antibiotic nucleus (whose solubility in the considered
pH range is the lowest possible), and partitioning of the sec-
ond reaction product (phenylacetic acid) between the two
phases. In addition to the reduction of technological steps,
the suggested process can improve the quality and the yield
of the reaction product since several factors leading to the
degradation of ꢀ-lactam ring are excluded. It is worth noting
that two-phase water-immiscible organic solvent systems have
so far only been evaluated for performing synthetic reactions
(6, 7), while the hydrolytic potential of these systems has
been deprived of proper attention. For practical implementa-
tion of this process in hydrolysis (e.g., in the countercurrent
Theoretical aspects
Thermodynamics of hydrolysis in a two-phase system at
low pH
Thermodynamic regularities of antibiotic hydrolysis in a
two-phase aqueous solution – water-immiscible organic sol-
vent system should take into account the following
physicochemical processes:
(i) Partition of acidic form of antibiotic (ABH) between
two phases:
[1]
ABHorg ꢁ ABH
with a partition constant of
[ABHorg
[ABH]
]
[2]
KP,ABH ꢂ
(ii) Dissociation of ABH in an aqueous phase (pK of the
carboxylic group of ꢀ-lactam antibiotics is close to 2.5):
[3]
ABH ꢁ ABꢃ + H+
with a dissociation constant of
[Hꢄ ][ABꢃ ]
[4]
Ka,ABH ꢂ
[ABH]
(iii) Enzymatic hydrolysis in an aqueous phase:
[5]
ABꢃ + H2O ꢁ AH + Bꢃ
and the corresponding equilibrium constant:
© 2002 NRC Canada

Chilov and Švedas
701
[ABꢃ ]
[AH][Bꢃ ]
Concentration of a certain ionic species in an aqueous me-
dium can be expressed as
[6]
Ksynth,AB ꢂ
[20] [ABꢃ] = [AB]FAB
ꢃ
where AH is a protonated form of phenylacetic acid and B–
is an anionic form of antibiotic nucleus.
(iv) Partition of a protonated form of phenylacetic acid be-
tween two phases:
[21] [AH] = [A]FAH
where
1
[7]
AH ꢁ AHorg
[22]
[23]
FAB
ꢂ
[Hꢄ ]
1 ꢄ
1 ꢄ
(1 ꢄ ꢆ KP,ABH)
with a partition constant of
Ka,ABH
[AHorg
[AH]
]
[8]
KP,AH
ꢂ
1
FAH
ꢂ
[Hꢄ ]
(1 ꢄ ꢆ KP,AH
)
Ka,AH
(v) Dissociation of phenylacetic acid in an aqueous phase:
[9]
AH ꢁ Aꢃ + H+
with a dissociation constant (pK 4.3) of
[Hꢄ ][Aꢃ ]
which follow from the eqs. [2], [4], [8], and [10], respec-
tively (where ꢆ is the phase-volume ratio). If concentration
of the Zwitter-ionic form is confined by its solubility
M
[
BHsol
],
[10] Ka,AH
ꢂ
[AH]
Ka1,B
[Hꢄ ]
[24] [B] = [BHsMol
]
(vi) Protonation of an amino group of the antibiotic nu-
cleus (pK about 4.7):
then the thermodynamic equilibrium of hydrolysis can be
expressed in the following way:
[11] Bꢃ + H+ ꢁ BHM
with a dissociation constant of
[Bꢃ ][Hꢄ ]
(1ꢃꢅ)[AB]0 FAB
[25] Ksynth, AB
=
Ka1,B
ꢅ ꢇꢈ]0 FAH[BHsMol
]
[Hꢄ ]
[12] Ka2,BH
ꢂ
[BHM ]
The final expression for the reaction yield can be written
(vii) Protonation of a carboxylic group of the nucleus
(pK 2.5):
as
1
[26] ꢅ =
[13] BHM + H+ ꢁ BH2M
Ka1,B
FAH
[Hꢄ ] FAB
1 ꢄ Ksynth, AB [BHsMol
]
with a dissociation constant of the Zwitter-ionic form of
[BHM ][Hꢄ ]
Factors that influence the yield of hydrolysis
[14] Ka1,BH
ꢂ
[BHꢄ2 ]
As seen from the previous paragraph, the yield of the anti-
biotic nucleus is a complex parameter determined by the
thermodynamic constant of hydrolysis, the dissociation and
partition constants, and the solubility of the product. Using
eq. [26], we can estimate their impact on the outcome of the
process.
(viii) Precipitation of the Zwitter-ionic form of antibiotic
nucleus:
[15] BHM ꢁ BHꢄppt
The effectiveness of the hydrolysis in a two-phase system
under a given set of conditions does not depend on the initial
antibiotic concentration. Such a conclusion might seem
strange, especially if one takes into consideration that in the
industrial processing of penicillin, hydrolysis at pH 7–8 (its
starting concentration) does not exceed 0.25 M — because
of the thermodynamic limitations. A further increase of the
substrate concentration would lead to the decrease in the
yield of 6-APA (9). But this conclusion is only valid for ho-
mogeneous solutions. If precipitation of the product(s)
occurs, such a dependence on the initial substrate concentra-
tion would also disappear. From a practical point of view,
the larger the substrate concentration converted, the higher
the productivity of the process.
which should make its concentration in an aqueous phase
equal to its solubility:
[16] [BHM ] = [BHsMol
]
Bearing in mind the relations mentioned above, we can
trace the factors that determine the yield of antibiotic nu-
cleus in the corresponding hydrolysis. Reserving ꢅ for the
reaction yield, the total content of each component (i.e., con-
centrations of all ionic forms in the co-existing phases) can
be represented as
[17] [AB] = (1 – ꢅ)[AB]0
[18] [A] = ꢅ[AB]0
In evaluating the influence of different factors on the
degree of antibiotic conversion, it is important to note that
when some characteristics of the two-phase system (thermo-
[19] [B] = ꢅ[AB]0
© 2002 NRC Canada

702
Can. J. Chem. Vol. 80, 2002
Fig. 2. Dependence of 6-APA yield during Pen G hydrolysis in a
two-phase system, calculated with the eq. [18] on the basis of
constants given in Table 1. Dashed line ꢆ = 1; solid line ꢆ = 5;
dotted line ꢆ = 0.2.
dynamic constant of hydrolysis, dissociation and partition
constants of certain reaction components) are fixed, parame-
ters such as pH and phase volume ratio (ꢆ) represent extra
degrees of freedom, which must be taken into account when
looking for optimal processing conditions.
1.0
0.9
0.8
0.7
The pH dependence of a conversion is stipulated by the
presence of ionogenic groups capable of dissociation in a pH
range of 3–4. It follows from thermodynamic considerations
that hydrolysis in an alkaline medium is nearly irreversible.
Degradation of ꢀ-lactam rings at higher pH, however, limits
its application for the biocatalytic process in this pH range,
making the biocatalytic process optimal in a pH range of 7–
8. In a two-phase system, the situation is quite different. Re-
moval of both reaction products out of the reaction sphere
(i.e., precipitation of 6-APA and partition of phenylacetic
acid) promotes enzymatic hydrolysis, making the
biocatalytic process optimal in acidic medium. In a two-
phase system, a pH increase above pH 4.4 hampers extrac-
tion of phenylacetic acid into the organic phase, prevents 6-
APA precipitation, and, in contrast to homogeneous reac-
tions, does not improve hydrolysis. Analysis of eq. [26]
shows that with a pH decrease, antibiotic conversion ap-
proaches a definite value:
3.0
3.3
3.6
3.9
4.2
pH
to be quite acceptable. To illustrate this point, we first ana-
lyzed the dependence of the yield of 6-APA, calculated on
the basis of the model presented above (Fig. 2), using ther-
modynamic parameters from the literature (Table 1).
According to thermodynamic predictions, the yield of 6-
APA under pH 3–4 with a single hydrolytic step is close to
90%, which may be considered quite satisfactory. But one
should determine whether such yields are achievable in prac-
tice and if the kinetics of the thermodynamically controlled
hydrolysis of antibiotics under a given set of conditions is
achievable. The aim of this paper is to report the results of
Pen G and Ceph G hydrolysis in a two-phase aqueous solu-
tion – water-immiscible organic solvent system at acidic pH.
1
[27] ꢅ pHJ0
=
Ka1,B 1 ꢄ KP,ABH
ꢆ
1 ꢄ Ksynth,AB [BHsMol
]
Ka,AB 1 ꢄ KP,AH
ꢆ
It is quite clear that the optimal yield is determined both
by thermodynamic characteristics and phase volume ratio
(ꢆ). The influence of the latter parameter, however, is quite
limited. Generally, the values of the partition constants are
several tens in its magnitude, and hence under reasonable
values of ꢆ the value of KP ꢆ sufficiently exceeds the unity.
Taking this into consideration, eq. [27] may be reduced to
Materials and methods
1
[28] ꢅ pHJ0
=
Materials
Ka1,B KP,ABH
Ka,AB KP,AH
1 ꢄ Ksynth,AB [BHsMol
]
6-APA, 7-ADCA, Pen G-K salt, Ceph G, and penicillin
acylase from E. coli (ATCC 11105) were kind gifts from
DSM (the Netherlands). The concentration of the enzyme-
active sites was determined by titration with PMSF (15) and
found to be 3.2 × 10–4 M. PAA was from Aldrich, butyl ace-
tate from Fluka, acetonitrile (HPLC grade) from Kriochrom
(Russia), sodium dodecylsulfate, the components of the
buffer systems, and all other reagents were from Merck. In
all experiments MilliQ (Millipore) water was used.
Finally, within the set of thermodynamic parameters and
their combinations (which determine the yield), we can dis-
tinguish three basic parameters: (i) The product of the ther-
modynamic antibiotic synthesis constant and the solubility
of the Zwitter-ionic form of the nucleus (Ksynth,AB[BHsMol ]). It
should be emphasized that the impact of these two parame-
ters on the yield is reciprocal, i.e., any deterioration of the
thermodynamics of the hydrolysis could, in principle, be
compensated for by lowering the product solubility. (ii) The
difference in the pK of the amino group of antibiotic nuclei
and the carboxylic group of the substrate, where protonation
of the former leads to the domination of the Zwitter-ionic
form (low solubility), while deprotonation of the latter as-
sists in the extraction of the substrate into aqueous phase.
(iii) The ratio of the partition constants of antibiotic and the
phenylacetic acid, where partition of the substrate into the
aqueous phase and product withdrawal to the organic phase
favors conversion.
Enzymatic hydrolysis in a two-phase system
Each experiment was performed in a series of microtubes
(Eppendorf) in a shaker at 20°C. The total reaction volume
was 500 L. A stock solution of Pen G (0.1 M) was pre-
pared by dissolving the corresponding amount of Pen G–K
salt in water. Equal volumes of BuAc and Pen G stock solu-
tions (250 L of each) were placed in a microtube and the
pH was adjusted to a required value with phosphoric acid
(less than 5 L). At cephalosporin hydrolysis, a definite
amount of Ceph G (acidic form) was placed in 250 L of
BuAc, the suspension was then treated in an ultrasonic bath
(Branson 2510) for 10 min to dissolve the substrate, the re-
quired volume of water was added, the system was stirred in
the centrifuge, and the pH was adjusted with a KOH solution.
Of particular interest is determining the degree of conver-
sion under given conditions. This is especially intriguing,
since biocatalytic hydrolysis at pH 7.5–8 was already shown
© 2002 NRC Canada

Chilov and Švedas
703
Fig. 3. Integral curves of Pen G (a) and Ceph G (b) hydrolysis
Table 1. Thermodynamic parameters of the processes involved in
in a two-phase system. (a) pH 3.8, 10 M PA; (b) pH 3.5,
10 M PA; (᭿), antibiotic (᭹), antibiotic nucleus.
antibiotic hydrolysis in a two-phase system under acidic pH.
Pen G Ceph G PAA 6-APA 7-ADCA
0.10
Ksynth (M–1)
pKCOOH
680a
2.5b
—
47e
—
n.d.
2.5b
—
n.d.
—
—
—
—
(a)
4.3b
—
2.6c
4.7c
0
2.2d
4.9d
0
0.08
0.06
0.04
0.02
0.00
K
p
NH2
28e
—
KP
Solubility (M)
0.01c
0.001d
aValues from ref. 10.
bValues from ref. 11.
cValues from ref. 12.
dValues from ref. 13.
eValues from ref. 14.
Reactions were started by adding 7.8 L of stock PA solu-
tion.
0
30
60
90
120
150
180
0.10
0.08
0.06
0.04
0.02
0.00
Reactions in close-to-equilibrium systems
(b)
Reaction conditions (e.g., total reaction volume, tempera-
ture, phase volume ratio) were similar to those described
above. Stock solutions were prepared by dissolving definite
amounts of antibiotic and phenylacetic acid in water; after
mixing with BuAc, a required amount of solid 6-APA (or 7-
ADCA) was added, and the pH was adjusted with phospho-
ric acid.
Analysis
To analyze the total content of both the aqueous and or-
ganic phases, 10 mL of the stop-reagent (50% of HPLC
eluent (v/v) and acetonitrile) was added to the reaction mix-
ture (500 L) at a definite moment. An aliquot of the ob-
tained solution was diluted with the eluent [potassium
phosphate buffer (0.68 g L–1, pH 3.0), sodium dodecylsulfate
(100 mg L–1), and CH3CN (24%, v/v)] and analyzed by
HPLC using a Waters system with a Chrompack Nucleosil-
C18 4.6 × 250 mm column, Waters Lambda-Max Model 481
detector at 210 nm, and a Waters 6000A pump with a flow
of 1 mL min–1. Once the 6-APA concentration in an aqueous
phase was determined, the reaction mixture was centrifuged
for 1 min at 13 200 rpm (Eppendorf 5415D), and then 10 L
of the aqueous phase were added to 1 mL of the stop-
reagent. After dilution with the eluent, an aliquot was sub-
jected to HPLC analysis.
0
30
60
90
120
150
180
t (min)
namic evaluation of the new approach, and hence, did not
apply standard techniques for PA stabilization (16, 17).
Penicillin degradation was studied by Reschke and
Schugerl (1), who observed lower stability of penicillin in an
aqueous phase under pH 3–4 and no degradation in the or-
ganic phase. Based on their experimental data and account-
ing for the partition of antibiotic in a two-phase system
(Table 1), we determined the chance of penicillin degrada-
tion at the stated experimental conditions to be less than 1%.
In fact, we did not observe Pen G decomposition within an
hour in a two-phase system at pH 3.5. Since Ceph G is
known to be more stable in an acidic medium, there was no
danger of decomposition.
Results and discussion
Typical experimental integral curves of penicillin acylase-
catalyzed ꢀ-lactam antibiotic hydrolysis in a two-phase sys-
tem are given in Fig. 3. Effective enzymatic hydrolysis of
both biosynthetic antibiotics Pen G and Ceph G took place
at pH 3.8, followed by precipitation of corresponding antibi-
otic nuclei, which is in good agreement with the preliminary
thermodynamic evaluation.
Withdrawal of both reaction products out of reaction
sphere, i.e., precipitation of 6-APA (or 7-ADCA) and extrac-
tion of phenylacetic acid into the organic phase, is favorable
from both a thermodynamic and a kinetic point of view. The
pH dynamics during the course of hydrolysis must, however,
being taken into consideration. Starting from the initial pH
(3–4), there was no significant change in pH over the course
Hydrolysis of antibiotics
Antibiotic hydrolysis in a two-phase aqueous solution –
water-immiscible organic solvent system could be compli-
cated by several factors, with the most negative ones being
enzyme inactivation in the organic phase at low pH and ꢀ-
lactam degradation.
The presence of a water-immiscible organic phase did not
influence the activity or stability of native penicillin acylase:
the loss of activity during the characteristic time of reaction
(1–3 h) did not exceed 10%, which is quite acceptable for
the thermodynamic evaluation of the system. Stabilized en-
zyme preparations should therefore be used in a preparative
biocatalytic process to meet the economic and technological
demands. In this study, we focused primarily on a thermody-
© 2002 NRC Canada

704
Can. J. Chem. Vol. 80, 2002
Fig. 4. Experimental dependence of conversion degrees in hydro-
lysis of 0.1 M antibiotic in a two-phase system. (᭢), 6-APA;
(ꢀ), 7-ADCA.
Fig. 5. Supersaturation at antibiotic nucleus formation in the
course of Pen G (a) and Ceph G (b) hydrolysis. (᭿), total anti-
biotic nucleus content; (᭹), concentration in aqueous phase;
(……), saturation level.
1.0
0.9
0.8
0.7
0.6
0.5
0.20
(a)
0.16
0.12
0.08
0.04
0.00
3.0
3.2
3.4
3.6
3.8
4.0
pH
0
50
100
150
200
250
of the reaction due to the balance of the dissociation and
partition constants, which prevented formation of unwanted
inorganic salts in a reaction mixture.
0.100
0.080
0.060
0.040
0.020
(b)
Experimental antibiotic nuclei yield dependence on pH
Hydrolysis of Pen G and Ceph G in a two-phase system
(water–BuAc, 1:1) at pH 3–4 was carried out and the pH de-
pendencies of the antibiotic conversion were obtained
(Fig. 4).
In a chosen interval, the yield of hydrolysis (70% for 6-
APA and 90% for 7-ADCA) was weakly dependent on pH.
It should be noted that the yield of 6-APA evaluated on the
basis of reported parameters (Fig. 2) exceeds the experimen-
tal yields by 20%. The discrepancy could arise from slightly
different experimental conditions used in different laborato-
ries when determining the numerous constants used for the
evaluation (eq. [27] contains several parameters determined
by different scientists). For example, there is a remarkable
discrepancy for the Pen G synthesis constant: Švedas et al.
(3) reported a value of 4300 M–1, while Tewari and
Goldberg (10) determined the value to be 680 M–1. Bearing
this in mind, we tested the thermodynamic model presented
above, but initially sought to characterize the equilibrium
state during the course of antibiotic hydrolysis in a two-
phase system.
It is interesting to compare antibiotic hydrolysis under
studied conditions in a biphasic system with that in a
monophasic one at the same acidic pH and at traditionally
used alkaline conditions (pH 8). Pen G and Ceph G hydroly-
sis in a monophasic aqueous system at low pH is thermody-
namically unfavorable (see corresponding values of
equilibrium constants) and is not practical due to very low
product yield (less than 30%) as well as low conversion rates
due mainly to very strong product inhibition at these condi-
tions (data not shown). Thus, in comparison with the corre-
sponding monophasic aqueous system, the biphasic system
(where both reaction products are removed out of the reac-
tion sphere due to extraction of phenylacetic acid into the or-
ganic phase and precipitation of 6-APA or 7-ADCA) favors
the hydrolytic reaction with thermodynamic and kinetic fea-
sibility.
0.006
0.003
0.000
0
20
40
60
80
100 120 140 160 180
t (min)
product inhibition, and therefore is just 5 to 6 times slower.
The total turnover rate is good enough for preparative con-
version. Moreover, the conversion rate can be further in-
creased by the appropriate choice of “acidic” PA, since
enzymes of this family are remarkably different in their pH
dependencies of catalytic activity (18). Furthermore, the
overall antibiotic conversion in a suggested technological
process can be increased up to the stoichiometric value by
applying a continuous downstream processing scheme. Anti-
biotic extraction from a cultural broth can be coupled with
further conversion in the biphasic system, which allows for
downstreaming in the feed-countercurrent reactor, and thus,
the possibility of reaching conversions that are close to unity
(19). When comparing traditional hydrolysis with a putative
one, the overall balance should take into account factors
such as the decrease in the number of the technological
steps, the formation of waste inorganic salt during numerous
pH shifts in the traditional procedure, and the economic
gains. It should also be done on the experimental process at
the preparative scale.
Dependence of the antibiotic conversion on its initial
concentration
The basic qualitative feature of the thermodynamic model
of ꢀ-lactam antibiotic hydrolysis in a two-phase system with
precipitation of the antibiotic nucleus is the independence of
Compared with the traditional hydrolysis of Pen G and
Ceph G at pH 8, the suggested approach is subjected to less
© 2002 NRC Canada

Chilov and Švedas
705
Fig. 7. Dependence of thermodynamically controlled antibiotic
conversion degree in a two-phase system. (᭢), 6-APA; (ꢀ), 7-
ADCA.
Fig. 6. Reaction in a two-phase system under close-to-
equilibrium conditions. (a) (ꢀ), 87 mM 6-APA, 87 mM PAA,
13 mM Pen G; (᭿), 89 mM 6-APA, 89 mM PAA, 11 mM Pen
G; (᭹), 95 mM 6-APA, 95 mM PAA, 5 mM Pen G; (b) (ꢀ),
98 mM 7-ADCA, 98 mM PAA, 2 mM Ceph G; (᭿), 99 mM 7-
ADCA, 99 mM PAA, 1 mM Ceph G; (᭹), 100 mM 7-ADCA,
100 mM PAA, 0 mM Ceph G.
1.0
0.9
0.8
0.7
0.6
0.5
0.014
(a)
0.012
0.010
0.008
0.006
0.004
3.0
3.2
3.4
3.6
3.8
4.0
4.2
pH
and supersaturation dissipates within several minutes, but
this is not the case in this particular system. The process of
antibiotic nuclei crystallization is complicated by various
factors such as contact with the organic phase, presence of
other reagents (Pen G and PAA), and the enzyme. In a sepa-
rate experiment, we observed that addition of 6-APA or 7-
ADCA crystals into the corresponding reaction mixture does
not influence precipitation, which emphasizes the peculiarity
of solid-phase formation under these experimental condi-
tions. We noted that 6-APA supersaturation dissipates very
slowly (tens of hours) and within the reasonable time inter-
val, one cannot expect the concentration of antibiotic nuclei
in aqueous phase to reach equilibrium. Despite a more sharp
precipitation for 7-ADCA, equilibrium solubility is also not
reached, and fourfold of oversaturation is left. Thus the sta-
bility of supersaturated solutions of antibiotic nuclei is most
probably responsible for the discrepancy between the experi-
mental and thermodynamic modeling presented above.
0
5
10
15
20
25
0.0020
0.0015
0.0010
0.0005
0.0000
(b)
0
10
20
30
40
t (min)
True equilibrium state of hydrolysis in a two-phase
system
the conversion on the initial substrate concentration. Experi-
mental study shows that this is a case for Ceph G hydrolysis,
where the yield of 7-ADCA does not depend on the Ceph G
concentration up to the 0.1 M (solubility limit of Ceph G un-
der these conditions). An increase in the Pen G concentra-
tion from 0.1 to 0.2 M, however, decreases the yield of 6-
APA from 67 to 57%. This fact pointed out one more dis-
crepancy between the thermodynamic model and the experi-
mental data. To solve this apparent contradiction, one might
change the model, but there was no sufficient reason for
such a change, and we wanted to be sure that the experimen-
tal data reflected a true equilibrium state. (Most of the
doubts were related to the precipitation of 6-APA.)
It is nevertheless important to know what yields could be
expected if the problems with precipitation of the nuclei
were solved. In other words, it is important to determine the
true equilibrium state of hydrolysis in a two-phase system.
For this purpose, we investigated the direction of the reac-
tion in systems with designed composition of reagents (Pen
G, 6-APA, PAA). By tracing the reaction direction it can be
concluded where the equilibrium, the state of no macro-
scopic reagent conversion, is situated. At the same time
there is no need to wait for the end of the reaction — it is
enough just to check the initial direction and then again with
each iteration step as the reaction approaches those concen-
trations that characterize the system at equilibrium. An anal-
ogous method is well known in mathematics as the method
of division by half for root estimation, where the sign of a
function is checked at each iteration step. If the sign
changes, the size of the interval, which contains the root or
the equilibrium (as in our case), is reduced. This method is
even more attractive because only the initial interval of reac-
tion is taken into account and thus supersaturation, reagent
degradation, and other dynamic effects can be omitted.
Dynamic supersaturation of 6-APA in the course of
reaction
Investigation of supersaturation effects was carried out by
comparing the composition of aqueous phase with that of
the whole system. Results are presented in Fig. 5.
Supersaturation takes place in both cases, and its scale is
impressive: 12-fold for 6-APA and 35-fold for 7-ADCA. It
therefore makes sense to consider these results in more de-
tail. Normally, supersaturated solutions are usually unstable
© 2002 NRC Canada

706
Can. J. Chem. Vol. 80, 2002
where [BHMeq] is the saturated (equilibrium) level and [BHMkin
is the supersaturated (kinetically stipulated) concentration;
ꢅeq and ꢅkin are the respective yields. Thus, on the basis of
supersaturations observed, one can recalculate expected
yields with the use of eq. [29]. Results of these calculations
are presented in Fig. 8.
As can be seen, after formal elimination of supersatu-
ration, the expected yields approximate the equilibrium val-
ues with good accuracy. This fact highlights that both the
experimental results and the theoretical conceptions intrinsi-
cally agree.
]
Fig. 8. Expected degrees of conversion in a two-phase system af-
ter formal elimination of supersaturation in accordance with
eq. [21]. (᭢), 6-APA; (ꢀ), 7-ADCA.
1.0
0.9
0.8
0.7
0.6
0.5
Conclusion
A process of Pen G and Ceph G hydrolysis is studied in a
two-phase system (water–BuAc) at pH 3–4. The thermody-
namic model of the process is analyzed, incorporating the
influence of pH, phase volume ratio, initial substrate concen-
tration, and parameters of hydrolytic reaction (acid-base and
interphase equilibria). For effective hydrolysis, we empha-
size the crucial role of product withdrawal from the reaction
sphere in a two-phase system: extraction of the emerging
phenylacetic acid as well as precipitation of the antibiotic
nucleus during the course of hydrolysis result in high yields.
Experimental tests also revealed the formation of antibiotic-
nucleus supersaturated solutions. For these solutions, being
rather stable, the yield in hydrolytic reactions is smaller than
predicted by the thermodynamic model. To study the true
thermodynamic equilibrium state, a method is proposed
where the initial reagent concentrations in an experimental
system are set close to the equilibrium concentration values.
Depending on whether hydrolysis or synthesis takes place,
the interval of equilibrium concentrations is returned.
Supersaturation effect is shown to decrease 6-APA yield
from 90%, as it is set by thermodynamics, to 67%, and 7-
ADCA yield from 99 to 90%, correspondingly.
3.0
3.2
3.4
3.6
3.8
4.0
4.2
pH
Several typical dependencies are given in Fig. 6 to illustrate
this proposed approach.
As can be seen, loading the system with 0.087 M 6-APA,
0.087 M PAA, and 0.013 M Pen G (which corresponds to
87% yield) leads to further hydrolysis of the antibiotic, af-
fording equilibrium yields exceeding 87%. On the other
hand, designing a yield of 95% causes a synthetic reaction,
and thus the equilibrium value is less than 95%. Finally, in
the system containing 0.089 M 6-APA, 0.089 M PAA, and
0.011 M Pen G (89% yield), no macroscopic changes in the
concentrations were observed, which suggests a good ap-
proach to achieving equilibrium conditions.
A series of measurements under pH 3.2–4.1 was con-
ducted and the thermodynamic equilibrium state of Pen G
and Ceph G hydrolysis in a two-phase system (water–BuAc)
was found. Results are presented in Fig. 7.
Correspondence of true equilibrium state and practical
conversions
Acknowledgments
In the case of Pen G, thermodynamically estimated yields
of hydrolysis (Fig. 2) are in good agreement with experi-
mental values obtained under close-to-equilibrium condi-
tions (Fig. 7). Moreover, the latter values also agreed with
the yields in non-equilibrium (as far as nucleus precipitation
is concerned) reactions. In fact, if the crystallization process
is considered relatively slow in comparison with the chemi-
cal reaction, the yield will be determined by eq. [26], where
the equilibrium solubility should be replaced by the current
concentration (determined by supersaturation) in the aque-
ous phase. Then, other thermodynamic parameters being equal,
the difference between thermodynamically and kinetically
controlled yields will be stipulated only by the ratio of satu-
rated and current concentrations according to eq. [26]. After
several transformations, we obtain the following parametric
relation between the conversion degrees and the ratio of con-
centrations of precipitating component:
This work was supported by the Russian Foundation for
Basic Research (Grants 00–04–48658 and 01–04–06616).
We thank T. Van der Does for fruitful discussions and DSM
for donated reagents.
References
1. M. Reschke and K. Schugerl. Chem. Eng. J. 28, B1 (1984).
2. P.S. Nys, L.M. Elizarovskaya, N.N. Shellenberg, and E.M.
Savitskaya. Antibiotiki (Rus), 24, 895 (1979).
3. V.K. Švedas, A.L. Margolin, and I.V. Berezin. Enzyme
Microb. Technol. 2, 138 (1980).
4. I.V. Berezin, A.A. Klyosov, A.L. Margolin, P.S. Nys, E.M.
Savitskaya, and V.K. Švedas. Antibiotiki (Rus), 21, 411 (1976).
5. I.V. Berezin, A.A. Klyosov, V.K. Švedas, P.S. Nys, and E.M.
Savitskaya. Antibiotiki (Rus), 19, 880 (1974).
6. J.B. Jones. Tetrahedron, 42, 3351 (1986).
1
7. Y.L. Khmelnitsky, I.N. Zharinova, I.V. Berezin, A.V. Levashov,
and K. Martinek. Ann. N.Y. Acad. Sci. 501, 161 (1987).
8. A.J.J. Straathof, M.B. Diender, J.L. den Hollander, T. van der
Does, L.A.M. van der Wielen, and J.J. Heijnen. In Integration
of biotransformation and downstream processing applied to
ꢃ1
[BHMeq
[BHMkin
]
ꢅeq
1
[29]
ꢂ
]
ꢃ1
ꢅkin
© 2002 NRC Canada

Chilov and Švedas
707
ꢀ-lactam antibiotic synthesis. Abstracts. 5th Int. Symp. Biocat.
Biotrans. Darmstadt. 2001. p. 151.
List of symbols
9. V.K. Švedas. In Advances in science and technology. Ser.
Biotechnol. Edited by P.S. Nys and Yu.V. Mitin. VINITI Press.
Moscow. 1988.
10. Y.B. Tewari and R.N. Goldberg. Biophys. Chem. 29, 245 (1988).
11. M.B. Diender, A.J.J. Straathof, and J.J. Heijnen. Biocatal.
Biotransform. 16, 275 (1998).
12. B.P. Bruns, E.M. Savitskaya, N.N. Shellenberg, G.S. Libinson,
T.S. Kolygina, and E.N. Druzhinina. Antibiotiki (Rus), 7, 440
(1962).
13. P.S. Nys, E.V. Koltsova, N.N. Shellenberg, E.M. Savitskaya,
and M.M. Levitov. Antibiotiki (Rus), 21, 391 (1976).
14. M. Reschke and K. Schugerl. Chem. Eng. J. 28, B11 (1984).
15. V.K. Švedas, A.L. Margolin, S.F. Sherstiuk, A.A. Klyosov, and
I.V. Berezin. Bioorg. Khim. 3, 546 (1977).
ꢆ Phase volume ratio (Vorg /Vaq)
C(X) Cconcentration of species X (mol L–1
)
C0(X) Initial concentration of X (mol L–1
F(X) Mol fraction of species X
)
Ka Acidity dissociation constant (mol L–1
)
pK Negative logarithm of acidity dissociation constant
KP Partitioning constant
Ksynth Equilibrium synthesis constant (mol–1 L)
V Volume (L)
ꢅ Yield
Sub- or superscript
app Apparent
aq Aqueous
eq Equilibrium
kin Kinetic
16. G. Alvaro, R. Fernandez-Lafuente, R.M. Blanco, and J.M.
Guisan. Appl. Biochem. Biotechnol. 26, 181 (1990).
17. J. Masarova, D. Mislovicova, P. Gemeiner, and E. Michalkova.
Biotechnol. Appl. Biochem. 34, 127 (2001).
org Organic
sol Solubility
ppt Precipitate
18. V. Švedas, D. Guranda, L. van Langen, F. van Rantwijk, and
R. Sheldon. FEBS Lett. 417, 414 (1997).
19. L.A.M. van der Wielen, K. Ch, and A.M. Luyben. Curr. Opin.
Biotechnol. 3, 130 (1992).
© 2002 NRC Canada