A green-by-design bioprocess for l-carnosine production integrating enzymatic synthesis with membrane separation

Article abstract of DOI:10.1039/c9cy01622h

l-Carnosine (l-Car, β-alanyl-l-histidine) is a bioactive dipeptide with important physiological functions. Direct coupling of unprotected β-Ala (β-alanine) with l-His (l-histidine) mediated by an enzyme is a promising method for l-Car synthesis. In this study, a new recombinant dipeptidase (SmPepD) from Serratia marcescens with a high synthetic activity toward l-Car was identified by a genome mining approach and successfully expressed in Escherichia coli. Divalent metal ions strongly promoted the synthetic activity of SmPepD, with up to 21.7-fold increase of activity in the presence of 0.1 mM MnCl₂. Higher temperature, lower pH and increasing substrate loadings facilitated the l-Car synthesis. Pilot biocatalytic syntheses of l-Car were performed comparatively in batch and continuous modes. In the continuous process, an ultra-filtration membrane reactor with a working volume of 5 L was employed for catalyst retention. The dipeptidase, SmPepD, showed excellent operational stability without a significant decrease in space-time yield after 4 days. The specific yield of l-Car achieved was 105 g_Car g_catalyst^-1 by the continuous process and 30.1 g_Car g_catalyst^-1 by the batch process. A nanofiltration membrane was used to isolate the desired product l-Car from the reaction mixture by selectively removing the excess substrates, β-Ala and l-His. As a result, the final l-Car content was effectively enriched from 2.3% to above 95%, which gave l-Car in 99% purity after ethanol precipitation with a total yield of 60.2%. The recovered substrate mixture of β-Ala and l-His can be easily reused, which will enable the economically attractive and environmentally benign production of the dipeptide l-Car.

Full text of DOI:10.1039/c9cy01622h

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
A green-by-design bioprocess for L-carnosine
production integrating enzymatic synthesis with
membrane separation†
Cite this: Catal. Sci. Technol., 2019,
9, 5971
a
a
b
c
a
*
Dong-Ya Yin, ‡ Jiang Pan,‡ Jie Zhu, You-Yan Liu and Jian-He Xu
L-Carnosine (L-Car, β-alanyl-L-histidine) is a bioactive dipeptide with important physiological functions.
Direct coupling of unprotected β-Ala (β-alanine) with L-His (L-histidine) mediated by an enzyme is a
promising method for L-Car synthesis. In this study, a new recombinant dipeptidase (SmPepD) from Serratia
marcescens with a high synthetic activity toward L-Car was identified by a genome mining approach and
successfully expressed in Escherichia coli. Divalent metal ions strongly promoted the synthetic activity of
SmPepD, with up to 21.7-fold increase of activity in the presence of 0.1 mM MnCl2. Higher temperature,
lower pH and increasing substrate loadings facilitated the L-Car synthesis. Pilot biocatalytic syntheses of
L-Car were performed comparatively in batch and continuous modes. In the continuous process, an ultra-
filtration membrane reactor with a working volume of 5 L was employed for catalyst retention. The
dipeptidase, SmPepD, showed excellent operational stability without a significant decrease in space–time
−1
yield after 4 days. The specific yield of L-Car achieved was 105 gCar gcatalyst by the continuous process
−1
and 30.1 gCar gcatalyst by the batch process. A nanofiltration membrane was used to isolate the desired
Received 13th August 2019,
Accepted 23rd September 2019
product L-Car from the reaction mixture by selectively removing the excess substrates, β-Ala and L-His. As
a result, the final L-Car content was effectively enriched from 2.3% to above 95%, which gave L-Car in 99%
purity after ethanol precipitation with a total yield of 60.2%. The recovered substrate mixture of β-Ala and
L-His can be easily reused, which will enable the economically attractive and environmentally benign
production of the dipeptide L-Car.
DOI: 10.1039/c9cy01622h
rsc.li/catalysis
usually required, and the amino group of β-Ala needs to be
protected to avoid the formation of unwanted by-products
Introduction
L-Carnosine (L-Car) is a β,α-dipeptide composed of β-alanine
(β-Ala) and L-histidine (L-His) and is abundantly present in
mammalian skeletal muscles and brain tissue. As a natural
biologically active compound, L-Car has many important
physiological functions in vivo, including anti-oxidation, anti-
glycosylation, pH buffering and free hydroxyl radical
elimination through complexation with transition metal ions
such as zinc or copper.1 Therefore, L-Car is used widely in
medicine, cosmetics, food additives and other fields.
because of the poor specificity of the reaction.2 A commonly
used industrial route of L-Car synthesis is shown in
Scheme 1a.3 In this route, the amino group of β-Ala is
protected by using o-phthalic anhydride as a protective agent,
and after the reaction, the phthaloyl protecting group is
deprotected by hydrazine, a highly toxic reagent. Moreover,
the tedious reaction steps, severe reaction conditions and
high-energy consumption make this process environmentally
unfriendly, which should be replaced by other methods.
Enzymatic synthesis of dipeptides has become a popular
research topic because of the excellent selectivity and
environmental friendliness of enzymes.4 Lipases and
proteases have been widely used to synthesize various short
peptides in organic solvents.5 However, research on
enzymatic synthesis of L-Car mediated by a lipase or protease
has not been reported. Aminopeptidases are commonly used
for the synthesis of dipeptides.6 Many studies on
aminopeptidase-catalysed synthesis of L-Car have been
reported, using an activated derivative of β-Ala, such as
β-alaninamide (Scheme 1b)7 or β-alanine methyl ester
(Scheme 1c)8 as the acyl donor and unprotected L-His as the
Chemical synthesis of L-Car has been extensively reported.
In these methods, activation of the carboxyl group of β-Ala is
a State Key Laboratory of Bioreactor Engineering, East China University of Science
and Technology, Shanghai 200237, P.R. China. E-mail: jianhexu@ecust.edu.cn
b School of Chemistry and Molecular Engineering, East China University of Science
and Technology, Shanghai 200237, P.R. China
c College of Chemistry and Chemical Engineering, Guangxi University, Nanning
530004, P.R. China
† Electronic supplementary information (ESI) available: Experimental details,
enzyme characterisation, HPLC, 1H-NMR, 13C-NMR and optical rotation of the
product. See DOI: 10.1039/c9cy01622h
‡ These authors (D. Y. Y., J. P.) contributed equally to this work.
This journal is © The Royal Society of Chemistry 2019
Catal. Sci. Technol., 2019, 9, 5971–5978 | 5971

View Article Online
Paper
Catalysis Science & Technology
hydrolysis of L-Car to generate β-Ala and L-His. In 2010, Ueda
et al. synthesized L-Car from nonprotected β-Ala and L-His by
employing the yeast cell surface-displayed carnosinase CN1 as
a catalyst in an organic-aqueous biphasic system (Scheme 1e
),12 which provided a new enzymatic approach to synthesize
L-Car. However, in this report, the activity of the carnosinase
CN1 expressed on the surface of yeast cells was rather low. The
loading of lyophilized cells was as high as 80 g L−1, while the
titre of L-Car produced was only 4.5 mM after a 72 h reaction in
a system composed of 500 mM β-Ala and 100 mM L-His in
aqueous buffer and a triple volume of isooctane. The specific
yield was only 12.7 mgCar gcell−1. The low activity of the catalyst
severely limits the development of this process.
In addition, several L-Car hydrolases derived from
prokaryotes have also been reported, including peptidase V
(PepV) derived from Lactobacillus delbrueckii,13 β-alanyl
dipeptidase (BapA) derived from Pseudomonas sp.,14
aminoacyl-histidine dipeptidase (VaPepD) derived from Vibrio
alginolyticus15 and peptidase
D (EcPepD) derived from
Escherichia coli (E. coli).16 Among them, PepV and BapA
specifically catalyse the hydrolysis of β-alanyl dipeptides (β-Ala-
Xaa), whereas PepDs specifically hydrolyze aminoacyl Xaa-His.
In this paper, we identified a highly active dipeptidase
(SmPepD) from Serratia marcescens by a genome mining
approach using the reported sequences of prokaryotic L-Car
hydrolases as leading templates. The resultant dipeptidase
SmPepD chosen among 20 candidates was characterized
biochemically and synthetic reaction conditions were
optimized systematically. Under the optimized reaction
conditions, SmPepD could efficiently catalyse the coupling of
unprotected β-Ala with L-His, forming L-Car in one step with
very high productivity. An ultra-filtration membrane bioreactor
(UF-MBR) was applied to realize the continuous synthesis of
L-Car. By combining with nanofiltration technology to isolate
L-Car produced from the excess substrates β-Ala and L-His, the
final product L-Car was successfully harvested in kilogram scale
with a purity up to 99%.
Scheme 1 Typical L-Car synthesis by chemical or enzymatic methods.
acyl receptor. The aminopeptidase catalysed synthesis of
L-Car is a kinetically controlled reaction with various by-
products such as H-β-Ala-β-Ala-His-OH and poly-β-Ala-
OMeIJNH2) formed during the process, and enzymatic
hydrolysis of L-Car occurs with extended reaction times. The
reported highest titre of L-Car was only 19.3 mM in these
processes.8 Moreover, the relatively expensive L-His was used
in excessive loadings, making this process uneconomical.
In vivo, L-Car is synthesized from β-Ala and L-His by a
specific synthetase (EC 6.3.2.11) via an ATP-dependent
pathway, but the sequence of L-Car synthase from mammals
has not been reported.9 In 2018, Kino et al. obtained an
L-amino acid ligase YwfE from Bacillus subtilis, which
catalysed the synthesis of L-Car by coupling β-Ala with L-His
in an ATP-dependent manner, as observed for L-Car
synthetase (Scheme 1d).10 The product titre reached 11.4 mM
at a substrate loading of 12.5 mM when using the best
mutant N108E/H378K. The addition of an equal molar
amount of ATP made the process impractical.
Results and discussion
Identification and characterization of new dipeptidases for
L-Car synthesis
Genes of several reported specific L-Car dipeptidases, including
cytoplasmic non-specific carnosinase from Homo sapiens
(hCN2), PepV from Lactobacillus delbrueckii (LdPepV),
dipeptidase PepD from Vibrio alginolyticus (VaPepD) and E. coli
(EcPepD), were cloned and solubly expressed in E. coli. Human
carnosinase hCN1 was also cloned, but not successfully
expressed in a soluble form in either E. coli or Pichia pastoris.
The recombinant dipeptidases were purified and the activities
of hydrolysis and synthesis of L-Car were measured. As shown
in Table 1, three dipeptidases catalysed not only the hydrolysis
of L-Car to β-Ala and L-His, but also the reverse reaction, i.e.,
synthesis of L-Car. Among them, human hCN2 showed the
highest hydrolytic and synthetic activity, but further screening
of new dipeptidases with higher activity failed using hCN2 as a
Peptidases catalysing L-Car degradation have been reported.
Two mammalian-derived dipeptidases that catalyse the
hydrolysis of L-Car are identified as tissue-specific carnosinase
(CN1, EC 3.4.13.20) and cytoplasmic non-specific carnosinase
(CN2, EC 3.4.13.18).11 CN1 has a narrow substrate specificity
for aminoacyl histidine (Xaa-His), whereas CN2 is a broad-
spectrum dipeptidase. The carnosinase catalyses reversible
5972 | Catal. Sci. Technol., 2019, 9, 5971–5978
This journal is © The Royal Society of Chemistry 2019

View Article Online
Catalysis Science & Technology
Paper
Table 1 Activity of several L-Car dipeptidases
Sources
Enzymes
Hydrolytic act. (μmol min−1 mg−1
)
Synthetic act. (μmol min−1 g−1
)
Homo sapiens
Vibrio alginolyticus
Escherichia coli
Enterobacter
Erwinia tasmaniensis
Serratia marcescens
hCN2
1417 56
62.4 4.8
15.0 4.9
6.7 1.7
12.2 5.1
15.7 1.0
VaPepD
EcPepD
EnPepD
EtPepD
SmPepD
144
179 20
240
7
5
224 13
2245 249
107
8
The substrate solutions were adjusted with a sodium hydroxide (NaOH) aqueous solution to pH 8.0. Conversion was maintained within 10–20%.
leading template. Among the prokaryotic L-Car dipeptidases,
EcPepD showed a higher hydrolytic activity, whereas VaPepD
sites and two substrate-binding sites, as well as two zinc ions
bound in the active center.17 Some metal ions, especially
afforded
a
higher synthetic activity, indicating that the
transition divalent metal ions, such as Co2+, Ni2+ and Mn2+
,
activities toward L-Car hydrolysis may not be completely
consistent with their synthetic activities.
have a strong stimulation effect on the hydrolytic activity of
some aminoacyl-histidine dipeptidases.15,18 These metal ions
are predicted to also promote L-Car synthesis catalysed by
SmPepD. Thus, the effect of metal ions and EDTA as a metal
chelating agent on the L-Car synthesis activity of SmPepD was
examined. As shown in Table 3, transition divalent metal
ions and alkaline earth metal ions, such as Mn2+, Co2+, Ni2+,
Ca2+ and Mg2+, stimulated the synthesis activity of SmPepD,
Using the amino acid sequence of the prokaryotic
dipeptidases as a guide, new dipeptidases with an obvious
activity toward L-Car synthesis were identified from the
genome database. Three more candidates (EnPepD, EtPepD
and SmPepD) with higher activities are shown in Table 1.
Among them, a dipeptidase SmPepD derived from Serratia
marcescens showed the highest synthetic activity for L-Car
synthesis, despite having only 62% identity to the leading
enzyme VaPepD. The hydrolytic and synthetic activities of
SmPepD toward L-Car were 2245 and 107 μmol min−1
gprotein−1, respectively, which are 14.6 and 6.1 times higher
when compared with those of VaPepD.
The kinetic parameters of recombinant SmPepD and
VaPepD catalysed coupling of β-Ala and L-His were
determined. As shown in Table 2, the kcat and Km of SmPepD
toward the substrate β-Ala are much higher than those of
VaPepD, while the catalytic efficiency constant (kcat/Km) of
SmPepD (3.05 min−1 M−1) toward β-Ala is slightly lower than
that of VaPepD (3.98 min−1 M−1). In contrast, the kcat of
SmPepD toward L-His is 8.1 times that of VaPepD, while the
Km of SmPepD toward L-His is only 1.4 times that of VaPepD.
As a result, the catalytic efficiency constant (kcat/Km) of
SmPepD (442 min−1 M−1) toward L-His is 6.4 fold higher than
that of VaPepD (69.4 min−1 M−1). Thus, SmPepD showed
better performance in the synthetic reaction when compared
with VaPepD because of the higher catalytic efficiency of
SmPepD toward L-His.
and the enzyme activity was highest in the presence of Mn2+
.
The reaction rate increased by one order of magnitude (10-
fold) with the addition of only 0.01 mM MnCl2 into the
reaction mixture (Table S4†). When added with 0.1 mM
MnCl2, SmPepD showed the highest activity with a 22.7-fold
increase in activity when compared with the reaction without
the addition of metal ions. Increasing the MnCl2
concentration further caused a decrease in the SmPepD
activity. Addition of Zn2+ did not promote the synthesis
activity, and no inhibition was observed either. The presence
of 0.1 mM Cu2+ caused a complete loss of SmPepD activity.
Addition of 1.0 mM EDTA also resulted in no enzyme activity.
In our work, although only synthetic activity was examined,
the activation effect of transition metal ions was consistent
with previous reports.15,18 Furthermore, SmpepD was proved
to be a metal-dependent dipeptidase, because the enzyme
activity of SmPepD was completely lost by adding 1.0 mM
EDTA, a metal chelating agent. Menard et al. speculated that
the role of metal ions might be the stabilization of the
bridging water molecule, resulting in formation of hydroxide
ions.19 However, the varied binding energy of metal ions with
different ligands (side chains) of SmPepD may result in
different catalytic performances. The rearrangement of
The reported dipeptidase VaPepD belongs to the M20
metal peptidase subfamily (M20C), with five metal-binding
Table 2 Kinetic parameters of VaPepD and SmPepD for L-Car synthesis
β-Alaa
L-Hisb
Enzyme
Km [mM]
kcat [min−1
]
kcat/Km [min−1 M−1
]
Km [mM]
kcat [min−1
]
kcat/Km [min−1 M−1
]
VaPepD
SmPepD
225 57
2580 230
0.90 0.03
6.36 0.77
3.98
3.05
8.8 1.6
12.4 3.4
0.60 0.07
5.46 0.42
69.4
442.4
Synthetic activities were determined in 0.2 mL Tris-HCl (50 mM, pH 8.0) at 30 °C and 1000 rpm. Various substrate concentrations and
a
b
appropriate purified SmPepD were used. Conversion was controlled within 10–15%. 60 mM L-His with 0–4 M β-Ala. 4.0 M β-Ala with 0–80
mM L-His.
This journal is © The Royal Society of Chemistry 2019
Catal. Sci. Technol., 2019, 9, 5971–5978 | 5973

View Article Online
Paper
Catalysis Science & Technology
Table 3 Effects of metal ions and EDTA on the activity of dipeptidase
SmPepD
SmPepD showed excellent thermal stability with half-lives of
96 h and 84 h, respectively, whereas at 50 °C SmPepD was
relatively unstable with a half-life of only 6 h.
Metal
ions or
additives
Relative activitya [%]
0.1 mM
Temperature is an important factor that affects the
equilibrium of reversible reactions. The Keq constants of the
enzymatic L-Car synthesis between 30 and 50 °C were
measured. The enzymatic reactions at different temperatures
were compared. When 2 M β-Ala, 100 mM L-His and 300 U
L−1 SmPepD were loaded, the L-Car synthesis reactions at 30
°C and 40 °C reached their equilibria within 48 h, and the
equilibrium concentration of L-Car increased from 7.62 mM
at 30 °C to 8.48 mM at 40 °C. At 50 °C, however, due to the
instability of SmPepD at this temperature, the L-Car synthesis
reaction stopped within 6 h with the same enzyme dose,
resulting in only 4.21 mM L-Car (Fig. S13†). Supplements of
the biocatalyst were necessary to drive the reaction toward
equilibrium. The equilibrium concentration of L-Car at 50 °C
was estimated to be approximately 8.79 mM. Therefore,
increasing the temperature resulted in an increase in the Keq
calculated from 2.01 at 30 °C to 2.37 at 50 °C (Table S6†),
implying that the synthesis of L-Car is an endothermic
reaction. Taking into consideration the Keq of the reaction
and the thermostability of SmPepD, 40 °C appears to be the
most suitable reaction temperature.
Changing the pH affects not only the ionization state of
the amino acid substrates but also the equilibrium of the
L-Car synthesis. Therefore, we determined the SmPepD
activity and the equilibrium concentration of L-Car at pH
values between 7.0 and 9.0 at 40 °C (Table S7†). On the one
hand, as the pH increased, the equilibrium concentration of
L-Car decreased. On the other hand, SmPepD was more active
at pH 8.0 by 3.9 and 1.2 times when compared with the
activities observed at pH 7.0 and 9.0, respectively.
Considering both the synthetic efficiency of the biocatalyst
and the reaction Keq, pH 8.0 was chosen as the most suitable
reaction pH.
The solubility of β-Ala measured in water is very high,
reaching 579 g L−1 (ca. 6.5 M) at 40 °C, whereas that of
histidine is relatively low with a value of 34.7 g L−1 (ca. 224
mM). According to Le Chatelier's principle, increasing the
concentrations of β-Ala and L-His can promote effectively the
reaction equilibrium in the direction of the L-Car synthesis,
thereby increasing the titre of L-Car. Obviously, increasing
the β-Ala concentration was more advantageous for
promoting the L-Car synthesis, as shown in Table 4. When
the concentrations of β-Ala and L-His were increased to near
saturation (6500 mM and 180 mM, respectively), the
equilibrium concentration of L-Car increased up to 62.5 mM
(ca. 14.1 g L−1), which is 13.9-fold higher than the highest
recorded value.12
1.0 mM
Control
Mn2+
Ni2+
100
2274
251
228
173
149
126
2
2
100
2127
196
463
247
2
4
2
1
4
4
2
3
6
4
1
2
5
Co2+
Ca2+
Mg2+
Li+
175
147
Zn2+
Fe3+
101 10
97
100
7
7
n.d.b
n.d.b
n.d.b
EDTA
25
Cu2+
n.d.b
Synthetic activities were determined in 0.2 mL Tris-HCl (50 mM, pH
8.0) containing 2.0 M β-Ala, 100 mM L-His and 0.5 mg mL−1 purified
SmPepD with or without the metal ions. Reaction conditions: 30 °C
a
and 1000 rpm. The activity measured without any additive was
taken as 100%. b n.d.: not detected.
ligands in the active site of SmPepD, which is ideal for Mn2+
but less optimal for Cu2+, may lead to the weaker binding of
Cu2+ and the poorer enzymatic activity. Nevertheless,
thorough understanding of the differences in binding and
synthetic performance due to the binding of various metal
ions with active-site residues still requires further studies of
crystal structures and all-atom molecular dynamics
simulations of the enzyme.
Optimization of enzymatic L-Car synthesis
Dipeptidase catalysed direct coupling of β-Ala and L-His is a
reversible reaction. In a reaction system containing Tris-HCl
buffer (50 mM, pH 8.0), 2.0 M β-Ala, 100 mM L-His and 300 U
L−1 SmPepD incubated at 30 °C, the concentration of L-Car at
equilibrium was 7.52 mM and Keq was calculated to be 1.98.
The use of Tris-HCl salt may be unnecessary because the
high concentrations of β-Ala and L-His used, with isoelectric
points of 6.9 and 7.6, respectively, afford a suitable buffering
effect at the reaction pH (8.0). To test this, a neat aqueous
phase system without any buffering salt gave an enzymatic
reaction process that was close to the system containing Tris-
HCl, with an equilibrium concentration of 7.62 mM L-Car
(Table S5†).
Temperature and pH effects on the SmPepD activity were
also examined. SmPepD showed the highest activity at 50 °C,
which is 6.7 times higher than the activity at 30 °C (Fig. S9†).
The enzyme was activated in an alkaline environment with
the highest activity observed at pH 8.0 (Fig. S10†). The
thermostability of SmPepD was measured. Surprisingly,
strong activation was observed when SmPepD was incubated
at 30–50 °C. In the initial period of incubation, the residual
activity of SmPepD was significantly higher than its initial
value, and later the residual activity decreased with the
inactivation of SmPepD (Fig. S11†). At 30 °C and 40 °C,
The effect of Mn2+ addition on the progress of L-Car
synthesis was investigated because Mn2+ significantly improves
the SmPepD activity. With the addition of 0.1 mM MnCl2, the
enzymatic reaction of the L-Car synthesis reached equilibrium
in only 2 h under conditions of 2 M β-Ala, 100 mM L-His, 75 U
L−1 SmPepD (ca. 0.5 g L−1 lyophilized cell free extract), pH 8.0
5974 | Catal. Sci. Technol., 2019, 9, 5971–5978
This journal is © The Royal Society of Chemistry 2019

View Article Online
Catalysis Science & Technology
Paper
Table 4 L-Car production at various substrate concentrations
initiated by adding 5 g lyophilized cell free extract (ca. 750 U)
and 1.0 mmol MnCl2, and continued by agitating at 40 °C
and 200 rpm. After 6 h of reaction, the L-Car production
Molar ratio of
β-Ala to L-His
β-Ala [mM]
L-His [mM]
L-Car [mM]
reached 66.7 mM, giving a space–time yield of 60.3 g L−1 d−1
1 : 2
1 : 1
2 : 1
5 : 1
10 : 1
20 : 1
10 : 1
Saturated
50
100
200
100
100
100
100
100
100
200
180
0.19 0.01
0.43 0.12
0.90 0.17
2.19 0.07
4.40 0.02
8.48 0.07
16.3 0.7
62.5 0.5
−1
and a specific productivity of 30.1 gCar gcatalyst
.
Since β-Ala, L-His and L-Car are all water-soluble, it is easy
to separate the enzyme SmPepD from the mixture of the
product and substrates by using an ultrafiltration (UF)
membrane. Considering the high stability of the dipeptidase
SmPepD, the UF-MBR was used for the continuous
production of L-Car. A batch reaction was carried out initially,
and when the L-Car concentration was higher than 50 mM,
the continuous reaction was initiated through continuously
500
1000
2000
2000
6500
Reactions were performed in a 0.2 mL system containing varied
concentrations of substrates using a 1500 U L−1 lyophilized cell free
extract of SmPepD. Reaction conditions: 40 °C and 1000 rpm for
24 h.
feeding the saturated substrate solution at
a relatively
constant flow rate (ca. 0.5 L h−1), and the effluent was
collected at the same flow rate. The reaction lasted for 4 d. In
the continuous reaction phase (Fig. 2, phase II), SmPepD
showed high operational stability with a nearly constant
reaction rate of about 25 mmol h−1 L-Car. The excellent
stability of SmPepD in the continuous process may be
attributed to the stabilizing effect of the amino acids and
dipeptide on the enzyme. The average space–time yield
and 40 °C, while under similar conditions without MnCl2, the
titre of L-Car after an 8 h reaction was only 0.9 mM (Fig. 1).
Compared with the reaction without Mn2+, the SmPepD activity
in the practical synthesis increased by as much as 117 times
upon the addition of 0.1 mM MnCl2. Furthermore, when the
substrate loadings were increased to saturating concentrations,
the reaction at 40 °C reached equilibrium in 8 h with 0.1 mM
MnCl2 and the same dose of catalyst (0.5 g L−1 lyophilized cell
free extract) (Fig. S14†).
achieved during the continuous reaction was 34.9 g L−1 d−1
−1
and the resulting specific productivity was 105 gCar gcatalyst
.
Since the SmPepD catalysed synthesis of L-Car is a
reversible reaction, the relative weight content of L-Car in the
continuous reaction effluent was only 2.3% (w/w). It is
difficult to separate L-Car from the excess substrates β-Ala
and L-His because of their similar properties, including the
very close isoelectric points. Membrane technology is an
efficient method for product separation. A nanofiltration
(NF) membrane was used for the separation and refining of
L-Car. The molecular weights of β-Ala, L-His and L-Car are 89,
155 and 226 Da, respectively, so an NF membrane with a
molecular weight cut-off (MWCO) of 150 Da was selected.
The retention factors (R) of these three compounds
Kilogram-scale preparation of L-Car
A batch reaction for the L-Car synthesis was carried out in a
15 L stirred glass-tank reactor (working volume: 10 L) with a
thermostat water jacket, which was operated under the
optimized reaction conditions. Substrates β-Ala and L-His
were added excessively to ensure their saturation state during
the reaction. The pH of the reaction mixture was around 7.5
and not adjusted to simplify the reaction as well as the
subsequent product purification procedure. The reaction was
Fig. 1 Reaction process of L-Car synthesis with and without Mn2+
.
Fig. 2 Progress curves of the L-Car synthesis reaction in a continuous
MBR. Phase I: batch reaction; phase II: continuous reaction. Symbols:
(●) L-Car produced in batch reaction; (◆) L-Car produced in
continuous reaction; ( ) reaction rate. The reaction rate is defined as
the amount of L-Car produced by SmPepD per hour.
Reactions were performed in 100 mL flasks with 10 mL mixtures
containing 2 M β-Ala, 100 mM L-His, and 75 U L−1 lyophilized cell-free
extract of SmPepD, pH 8.0, magnetically agitated at 40 °C and 600
rpm. Symbols: ( ) with 0.1 mM Mn2+ ions; (▲) without Mn2+ ions.
This journal is © The Royal Society of Chemistry 2019
Catal. Sci. Technol., 2019, 9, 5971–5978 | 5975

View Article Online
Paper
Catalysis Science & Technology
determined were 46.5%, 65.7% and 92.3%, respectively,
indicating significant differences. Therefore, β-Ala and L-His
were efficiently permeated, while L-Car was selectively
retained. After enrichment by the NF membrane, the relative
content of L-Car was increased to more than 95% (w/w) with
an isolation yield of 67.2% after the nanofiltration process.
The excess substrates and a minor amount of product in the
permeate during the NF membrane separation can be reused
easily in a subsequent reaction. The crude product harvested
after NF enrichment was further refined through
decolourisation with activated carbon and removal of
residual MnCl2 by zeolite powder adsorption. After
subsequent ethanol precipitation and drying, a purified
product of L-Car was afforded with a purity of 98.8%. To sum
up, 72.3 g of L-Car was harvested per 10 L of the reaction
The recombinant E. coli was cultured in Luria-Bertani (LB)
medium supplemented with 50 μg mL−1 of kanamycin at 37
°C till the optical density at 600 nm reached 0.6–0.8.
Isopropyl-β-D-thiogalactopyranoside (IPTG) was then added to
the final concentration 0.2 mM and cells were cultivated at
16 °C for a further 24 h. The cells were collected and
disrupted by ultrasonification. After centrifugation, the cell
free extract was loaded onto a Ni-NTA column and eluted by
buffers containing different concentrations of NaCl and
imidazole. The purity of the recombinant enzymes in the
eluent was analysed via sodium dodecyl sulfate
polyacrylamide gel electrophoresis (SDS-PAGE). The purified
recombinant enzymes were collected for further assay.
Standard enzymatic assays for hydrolysis and synthesis of
L-Car
mixture, with
a total isolation yield of 60.2%. The
concentration of residual MnCl2 detected was lower than 2
Enzymatic hydrolysis of L-Car was performed in 0.2 ml Tris-
HCl buffer (50 mM, the pH was adjusted with a NaOH
solution to 8.0) containing 1.9 M β-Ala and 0.1 M L-Car. After
pre-incubation at 30 °C, the reaction was started by adding
adequate amounts of purified enzymes and then agitated at
30 °C and 1000 rpm for 5 min. The reaction was quenched
by mixing 40 μL of the reaction solution with a 760 μL HClO4
solution (0.1 M, in water) and the concentration of L-His
formed was analysed by HPLC. One unit of the enzyme was
defined as the amount of enzyme that hydrolysed 1.0 μmol
L-Car per minute under the above conditions.
Enzymatic synthesis of L-Car was performed in 0.2 mL
Tris-HCl buffer (50 mM, the pH was adjusted with a NaOH
solution to 8.0) containing 2.0 M β-Ala and 0.1 M L-His. The
reaction was performed as described above for 20 min. After
quenching, the concentration of L-Car formed was analysed
by HPLC.
−1
mg kgproduct (ppm) and the specific rotation of the product
was +21.0° (c = 3.0, water). Lit.20 L-Car [α]D25 + 21.9 (c = 3.0,
water).
Conclusions
A highly active dipeptidase SmPepD was cloned from Serratia
marcescens by a genome mining approach. The enzyme can
efficiently catalyse the reverse synthesis of L-Car through
direct coupling of the unprotected β-Ala and L-His. To the
best of our knowledge, this is the first report using a
prokaryote-derived dipeptidase for the synthesis of the
dipeptide L-Car. The Mn2+ ion had a very strong stimulation
effect on the synthetic activity of SmPepD. A continuous
enzymatic reaction in an UF-MBR was realized, and L-Car of
high purity was harvested through NF membrane separation.
The whole process is very simple and clean, without the use
of any buffer salt and any toxic solvent (except water and
edible ethanol), and without any by-products, making the
process a highly efficient and green synthesis process for
L-Car production.
For determining the kinetic parameters, the initial rates of
the enzymatic L-Car synthesis at different molar ratios of
β-Ala to L-His were determined and fitted to the Michaelis–
Menten model using the software Origin 9.0.
Chemical assay
Experimental
Chemicals and materials
Samples were analysed using
a reversed-phase HPLC,
equipped with a chiral column of CROWNPAK® (+) (150 × 4
mm, 5 μm particle size), Daicel Chiral Technologies (China)
Co., Ltd. After equilibration with HClO4 solution (pH 1.0, in
water) at a column temperature of 25 °C, compounds β-Ala,
L-His and L-Car were separated with an isocratic flow of
HClO4 solution (pH 1.0) at a constant flow rate of 0.3 mL
min−1, detected at 210 nm UV and quantified according to an
extraterritorial method. The content of Mn in the product
was analysed with an HPLC-inductively coupled plasma mass
spectrometer (NexION 2000-(A-10)), PerkinElmer Enterprise
Management (Shanghai) Co., Ltd, and the experimental data
were provided by the Analytical and Testing Center of East
China University of Technology. The details for the retention
times of β-Ala, L-His and L-Car are reported in the ESI†
(paragraph 5).
All chemicals and reagents were commercially available and
of the highest purity.
UF membrane PT 1812 and NF membrane DK 1812 with
MWCOs of 10 000 Da and 150 Da were obtained from Suez
Water & Process Technologies (Wuxi) Co., Ltd (China).
E. coli BL21 (DE3) and plasmid pET-28a (+) were used for
DNA manipulation and protein expression.
Gene cloning, expression and purification of recombinant
enzymes
The genes encoding putative L-Car hydrolases were amplified
by the PCR using the corresponding genomic DNAs as the
template. The amplified DNA fragment was ligated into the
pET-28a (+) vector and transformed into E. coli BL21 (DE3).
5976 | Catal. Sci. Technol., 2019, 9, 5971–5978
This journal is © The Royal Society of Chemistry 2019

View Article Online
Catalysis Science & Technology
Paper
Calculation of the equilibrium constant
condensed by rotary evaporation to recover the ethanol. The
residual solution mixture (5 L) was separated by using an NF
membrane (DK 1812, MWCO of 150 Da). A volume-constant
pattern was employed during the NF process using water as a
leachate solvent at an osmosis rate of 5 L h−1, till the relative
mass content of L-Car increased to over 95%.
After nanofiltration, the retentate was collected and 3 g
L−1 activated carbon was added and stirred at 60 °C for 0.5 h
for decolourisation. After cooling, the mixture was filtered
and 10 g L−1 artificial zeolite powder was added into the
filtrate and stirred at 30 °C for 0.5 h to adsorb the undesired
Mn2+ ions. The mixture was filtered again and the filtrate was
vacuum concentrated till a solid sediment was observed.
Then a 10-fold volume of ethanol was added slowly into the
saturated solution followed by gentle stirring. The mixture
was filtered and the filter cake was dried to constant weight.
Enzymatic hydrolysis and synthesis of L-Car were performed
simultaneously in a 10 mL system that was magnetically stirred
at 600 rpm. Both reactions were conducted under the same
conditions except for the substrate concentration. For the
hydrolytic reaction, 1.9 M β-Ala and 100 mM L-Car were used,
whereas for the synthetic reaction, 2.0 M β-Ala and 100 mM
L-His were used. 0.1 g of lyophilized cell free extract (15 U) was
added to start the reaction, and extra catalyst was supplemented
if necessary, till the concentration of L-Car in the reactions of
both directions became identical, that is, the reaction reached
equilibrium. Concentrations of β-Ala, L-His and L-Car at
equilibrium were determined by HPLC, and the equilibrium
constant (Keq) was calculated according to eqn (1).
Keq. = CCar × CH O/CAla × CHis
(1)
2
Conflicts of interest
where CCar, CH O, CAla and CHis represent the concentrations
2
of L-Car, water, β-Ala and L-His, respectively, at reaction
equilibrium.
There are no conflicts to declare.
Acknowledgements
Batch reaction for the preparation of L-Car
This work was financially sponsored by the National Key
The batch reaction was performed in a 15 L glass-jacked
stirred tank reactor. In the reactor, 5.874 kg β-Ala (6.6 M) and
387 g L-His (250 mM) were added, and water was added to
dissolve the substrates β-Ala and L-His at 40 °C and 200 rpm,
till the volume of the reaction mixture reached 10 L. The
reaction was started by adding 5 g lyophilized cell free extract
of SmPepD (750 U) and 198 mg MnCl2 (1.0 mmol), stirred at
40 °C and 200 rpm continuously. The samples were
withdrawn intermittently for the detection of the L-Car
concentration.
Research
and
Development
Program
of
China
(2016YFA0204300), the National Natural Science Foundation
of China (No. 21536004, 21776085 & 21871085), and the
Fundamental Research Funds for the Central Universities
(222201714026 & WF1714026).
Notes and references
1 (a) O. I. Aruoma, B. Halliwell, M. J. Laughton, G. J. Quinlan
and J. M. Gutteridge, Biochem. J., 1989, 258, 617–620; (b) R.
Kohen, R. Misgav and I. Ginsburg, Free Radical Res.
Commun., 1991, 12, 179–185; (c) E. C. Smith, J. Physiol.,
1938, 92, 336–343; (d) G. R. Lenz and A. E. Martell,
Biochemistry, 1964, 3, 745–750; (e) R. P. Bonomo, V. Bruno,
E. Conte and G. Vecchio, Dalton Trans., 2003, 4406–4415; ( f )
S. Cuzzocrea, T. Genovese, M. Failla and C. Vancheri, Am. J.
Physiol., 2007, 292, 1095–1104.
Continuous flow reaction based on a UF membrane for the
preparation of L-Car
A
continuous reaction was performed in a membrane
filtration device equipped with the UF membrane PT 1812
(MWCO 10 000 Da), as shown in Scheme S1.† The reaction
volume was 5 L. A batch reaction was performed initially in a
mixture containing saturated β-Ala and L-His, 0.1 mM MnCl2
and lyophilized cell free extract of SmPepD (750 U). When the
L-Car concentration reached 50 mM, the continuous flow
reaction began by continuously feeding the saturated
solution of β-Ala and L-His (containing 0.1 mM MnCl2), and
discharging the reaction mixtures at the same space velocity
(0.5 L h−1). The whole reaction process lasted for 4 d.
2 (a) H. Schwarz and K. Arakawa, J. Am. Chem. Soc., 1959, 81,
5691–5695; (b) E. Dakdouki, H. Mohammad, D. Nadine and
A. Hiba, Int. J. Pept. Res. Ther., 2019, 25, 379–390.
3 R. A. Turner, J. Am. Chem. Soc., 1953, 75, 2388–2390.
4 (a) G. Qu, B. B. Liu, Y. Y. Jiang, Y. Nie, H. L. Yu and
Z. T. Sun, Bioresour. Bioprocess., 2019, 6, 18; (b) J.
Tramper, Biotechnol. Bioeng., 1996, 52, 290–295; (c) C.
Liang, Y. Nie, X. Mu and Y. Xu, Bioresour. Bioprocess.,
2018, 5, 33; (d) Z. N. You, Q. Chen, S. C. Shi, M. M.
Zheng, J. Pan, X. L. Qian, C. X. Li and J. H. Xu, ACS
Catal., 2019, 9, 466–473.
5 (a) P. Stehle, H. P. Bahsitta, B. Monter and P. Furst, Enzyme
Microb. Technol., 1990, 12, 56–60; (b) F. Bordusa, Chem. Rev.,
2002, 102, 4817–4867; (c) K. Yokozeki and S. Hara,
J. Biotechnol., 2005, 115, 211–220; (d) X. G. Li and L. T.
Kanerva, Org. Lett., 2006, 8, 5593–5596; (e) S. M. A. Salam,
Product isolation and purification
Portions of the continuous reaction mixture (50 L) were
combined with a relative mass content of only 2.3% L-Car.
Every 10 L of the reaction mixture was mixed with an equal
weight (11.7 kg) of ethanol. Major contents of β-Ala and L-His
were precipitated because of their relatively low solubility in
an ethanol solution. After filtration, the filtrate was
This journal is © The Royal Society of Chemistry 2019
Catal. Sci. Technol., 2019, 9, 5971–5978 | 5977

View Article Online
Paper
K. I. Kagawa, T. Matsubara and K. Kawashiro, Enzyme
Catalysis Science & Technology
2003, 278, 6521–6531; (b) H. Otani, N. Okumura, A.
Okumura and K. Nagai, J. Biochem., 2005, 137, 167–175.
12 C. Inaba, S. Higuchi, H. Morisaka, K. Kuroda and M. Ueda,
Appl. Microbiol. Biotechnol., 2010, 86, 1895–1902.
13 K. F. Vongerichten, J. R. Klein, H. Matern and R. Plapp,
Microbiology, 1994, 140, 2591–2600.
14 H. Komeda and Y. Asano, FEBS J., 2010, 272, 3075–3084.
15 T. Y. Wang, Y. C. Chen, L. W. Kao, C. Y. Chang, Y. H. Liu,
J. M. Feng and T. K. Wu, FEBS J., 2008, 275, 5007–5020.
16 N. Okumura, J. Tamura and T. Takao, Protein Sci., 2016, 25,
511–522.
17 (a) C. Y. Chang, Y. C. Hsieh, T. Y. Wang, Y. C. Chen, Y. K.
Wang, T. W. Chiang, Y. J. Chen, C. H. Chang, C. J. Chen and
T. K. Wu, J. Biol. Chem., 2010, 285, 39500–39510; (b) C. Y.
Chang, Y. C. Hsieh, T. Y. Wang, C. J. Chen and T. K. Wu,
Acta Crystallogr., Sect. F: Struct. Biol. Cryst. Commun.,
2009, 65, 216–218.
Microb. Technol., 2008, 43, 537–543.
6 (a) T. Heck, H. P. E. Kohler, M. Limbach, O. Flögel, D.
Seebach and B. Geueke, Chem. Biodiversity, 2007, 4,
2016–2030; (b) T. Heck, V. S. Makam, J. Lutz, L. M. Blank, A.
Schmid, D. Seebach, H. E. Kohler and B. Geueke, Adv. Synth.
Catal., 2010, 2, 407–415.
7 J. Heyland, N. Antweiler, J. Lutz, T. Heck, B. Geueke, H. E.
Kohler, L. M. Blank and A. Schmid, Microb. Biotechnol.,
2010, 3, 74–83.
8 R. Takeshita, Y. Takakura, S. Suzuki and K. Yokozeki, US
20110081678A1, 2011.
9 (a) K. Bauer and M. Schulz, Eur. J. Biochem., 1994, 219,
43–47; (b) G. D. Kalyankar and A. Meister, J. Biol. Chem.,
1959, 234, 3210–3218; (c) S. D. Skaper, S. Das and F. D.
Marshall, J. Neurochem., 1973, 21, 1429–1445; (d) R. E.
Winnick and T. Winnick, Biochim. Biophys. Acta, 1959, 31,
47–55.
18 A. Aoki, Y. Shibata, S. Okano, F. Maruyama, A. Amano, I.
Nakagawa and Y. Abiko, Microb. Pathog., 2012, 52, 17–24.
19 H. A. Lindner, A. Alary, L. I. Boju, T. Sulea and R. Menard,
Biochemistry, 2005, 44, 15645–15651.
20 M. S. Cherevin, Z. P. Zubreichuk and L. A. Popova, Russ. J.
Gen. Chem., 2007, 77, 1576–1579.
10 K. Kino, US 20180179511A1, 2018.
11 (a) M. Teufel, V. Saudek, J. P. Ledig, A. Bernhardt, S.
Boularand, A. Carreau, N. J. Cairns, C. Carter, D. J. Cowley,
D. Duverger, A. J. Ganzhorn, C. Guenet, B. Heintzelmann, V.
Laucher, C. Sauvage and T. Smirnova, J. Biol. Chem.,
5978 | Catal. Sci. Technol., 2019, 9, 5971–5978
This journal is © The Royal Society of Chemistry 2019