Fluorescence spectroscopy as a novel method for on-line analysis of biocatalytic C-C bond formations

Article abstract of DOI:10.1016/j.molcatb.2010.04.006

On-line analysis of bioprocesses is of increasing interest avoiding the time delay for off-line sample preparation and the following analyses via chromatographic methods. Moreover, continuous monitoring of the reaction components during chemo- or biocatalytic transformations provides a direct control of the process. Since productivity of the processes can be controlled simultaneously, on-line monitoring of the processes is attractive for industrial applications. The reliable in situ monitoring of biocatalyzed reactions has been a challenge where reactions run in aqueous solutions. Limited work has been published on the use of spectroscopic methods for on-line analysis of biocatalytic reactions up to now. However, in this communication two dimensional (2D)-fluorescence spectroscopy has been proved to be an effective tool for on-line monitoring of the carboligation reactions catalyzed by wild type benzoylformate decarboxylase (BFD) from Pseudomonas putida. BFD is a thiamine diphosphate (ThDP)-dependent enzyme that catalyzes the asymmetric C-C bond formation to (S)-2-hydroxypropiophenone ((S)-2-HPP) starting from benzaldehyde and acetaldehyde. The analysis of the fluorescence spectra was achieved by chemometric modeling performing principle component analysis (PCA) and partial least square (PLS) regression. The derived chemometric models were used for the validation of concentrations of yielded 2-HPP and the substrate benzaldehyde with low root mean square error of calibration (RMSEC).

Full text of DOI:10.1016/j.molcatb.2010.04.006

Journal of Molecular Catalysis B: Enzymatic 66 (2010) 124–129
Contents lists available at ScienceDirect
Journal of Molecular Catalysis B: Enzymatic
journal homepage: www.elsevier.com/locate/molcatb
Fluorescence spectroscopy as a novel method for on-line analysis of biocatalytic
C–C bond formations
Selin Karaa, Fabienne Antonb, Dörte Solleb, Markus Neumanna,
Bernd Hitzmannb, Thomas Scheperb, Andreas Liesea,∗
a Institute of Technical Biocatalysis, Hamburg University of Technology (TUHH), Denickestrasse 15, D-21073 Hamburg, Germany
b Institute of Technical Chemistry, Leibniz University of Hanover, Callinstrasse 3, D-30167 Hannover, Germany
a r t i c l e i n f o
a b s t r a c t
Article history:
On-line analysis of bioprocesses is of increasing interest avoiding the time delay for off-line sample
preparation and the following analyses via chromatographic methods. Moreover, continuous monitoring
of the reaction components during chemo- or biocatalytic transformations provides a direct control of
the process. Since productivity of the processes can be controlled simultaneously, on-line monitoring of
the processes is attractive for industrial applications. The reliable in situ monitoring of biocatalyzed reac-
tions has been a challenge where reactions run in aqueous solutions. Limited work has been published
on the use of spectroscopic methods for on-line analysis of biocatalytic reactions up to now. However, in
this communication two dimensional (2D)-fluorescence spectroscopy has been proved to be an effective
tool for on-line monitoring of the carboligation reactions catalyzed by wild type benzoylformate decar-
boxylase (BFD) from Pseudomonas putida. BFD is a thiamine diphosphate (ThDP)-dependent enzyme that
catalyzes the asymmetric C–C bond formation to (S)-2-hydroxypropiophenone ((S)-2-HPP) starting from
benzaldehyde and acetaldehyde. The analysis of the fluorescence spectra was achieved by chemometric
modeling performing principle component analysis (PCA) and partial least square (PLS) regression. The
derived chemometric models were used for the validation of concentrations of yielded 2-HPP and the
substrate benzaldehyde with low root mean square error of calibration (RMSEC).
Received 4 December 2009
Received in revised form 5 March 2010
Accepted 8 April 2010
Available online 14 May 2010
Keywords:
On-line monitoring
Fluorescence spectroscopy
Enzymatic catalysis
Benzoylformate decarboxylase
C–C bond formation
Chemometrics
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
monitored via UV measurements. The reliable in situ monitoring
of biocatalyzed reactions has been a challenge where reactions run
In recent years many procedures for on-line monitoring of
biotransformations have been reported, e.g., ultraviolet (UV) spec-
troscopy [1], fluorescence spectroscopy [2], bioluminescence [3],
flow calorimetry [4], dielectric spectroscopy [5], or Fourier trans-
formed infrared (FT-IR) spectroscopy [6]. With on-line analysis
methods reaction systems can be monitored faster and more
efficiently than with conventional methods. On-line analysis
techniques enable production control in order to enhance proce-
dures and processes, as well as for quality control. Thus, on-line
monitoring results in improved ecologic and economic process
management. Moreover, rapid enzyme screening is achieved by
colorimetric assays where the color formation can be continuously
in aqueous solutions. Since water is the dominant component and
the reacting species are at low concentrations, obtained spectra
are usually of poor quality in infrared (IR) spectroscopy. Among
the developed methods, two dimensional (2D)-fluorescence anal-
ysis has been successfully applied for the on-line monitoring of
cultivations where the NAD(P)H-pool inside living organisms is
followed. Via fluorescence measurements bioprocess parameters
can be derived and be controlled in order to achieve high biomass
yield [7]. In addition to cultivation applications, 2D-fluorescence
spectroscopy has also been applied for the analysis of enzymatic
catalyses. Applications are found to examine pseudoenantiomeric
reactions catalyzed by proteases and esterases [2(a,b)] and to
analyze the NADH-dependent biotransformations catalyzed by an
oxidoreductase [2(c)].
The BioView® sensor (DELTA, Denmark) was developed as
a robust instrument for 2D-fluorescence measurement of bio-
processes especially in an industrial environment [7]. The main
advantage is the absence of a direct contact between sensor and
medium which eliminates the risk of contamination. No sampling
is necessary to get a direct insight into the process itself. There-
fore, 2D-fluorescence spectroscopy is a promising analytic method
Abbreviations: BFD, benzoylformate decarboxylase; BSA, bovine serum albumin;
FAD, flavinadenine dinucleotide; FMN, flavin mononucleotide; HPP, hydroxypropio-
phenone; PCA, principal component analysis; PLS, partial least square; RFI, relative
fluorescence intensity; RMSEC, root mean square error of calibration; ThDP, thi-
amine diphosphate; Wt, wild type.
∗
Corresponding author. Tel.: +49 40428 78 3018; fax: +49 40428 78 2127.
E-mail address: liese@tuhh.de (A. Liese).
1381-1177/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcatb.2010.04.006

S. Kara et al. / Journal of Molecular Catalysis B: Enzymatic 66 (2010) 124–129
125
and kanamycin (0.05 g/L). The oxygen saturation was adjusted to
at least 40% by varying the stirrer speed and by the aeration with
compressed air and pure oxygen.
After 24 h the overexpression was induced by the addition of iso-
propyl ␤-d-1-thiogalactopyranoside (IPTG) to a final concentration
of 1.5 mM at an OD600 nm = 50. Temperature was decreased from
37 ◦C to 28 ◦C and 18 h after induction cells were harvested by cen-
trifugation (10,000 rpm for 30 min at 4 ◦C) and stored at −20 ◦C.
Purification of the wtBFD was performed by immobilized metal
affinity chromatography Ni-NTA column as previously describes
[10]. BFD was applied in biotransformations as freeze-dried prepa-
rations. Protein amounts in the solutions were determined by the
standard Bradford method [11] using bovine serum albumin (BSA)
as a standard.
Fig. 1. The synthesis of (S)-2-HPP from benzaldehyde and acetaldehyde catalyzed
by wtBFD.
for the on-line monitoring of bioprocesses including the enzymatic
reactions which are performed under harsh conditions, especially
under high pressure [2(a,b)]. Our future goal is to transfer the
applicability of 2D-fluorescence spectroscopy into the biocatalyses
which are performed under high pressure conditions.
2.2.2. Biotransformations catalyzed by wtBFD
The bi-functional nature (keto- and hydroxy groups) and the
presence of a stereocentre make the chiral 2-hydroxyketones
important as building blocks for several pharmaceuticals. Thus,
synthesis of these key intermediates is of high interest. Most
importantly, these crucial compounds can be synthesized via
biocatalyzed asymmetric C–C bond formations with high enan-
tiomeric excesses (ee) [8]. One of the enzymes applied in the
synthesis of these chiral intermediates is benzoylformate decar-
boxylase from Pseudomonas putida (BFD, EC 4.1.1.7). BFD is a
homotetrameric thiamine diphosphate (ThDP)-dependent enzyme
which catalyzes the synthesis of chiral 2-hydroxyketones accept-
ing a broad range of aldehydes as substrates. Besides ThDP,
divalent metal cations, preferable Mg2+ is required as a second
cofactor in biotransformations. Wild type (wt) BFD is catalyz-
ing the carboligation of benzaldehyde and acetaldehyde yielding
2-hydroxypropiophenone (HPP) with ee of 92% (S) (Fig. 1) [9]. Syn-
thesis of 2-HPP and its derivatives has been intensively studied
[9(a,b)].
Reactions were carried out in a steel reactor with a total vol-
ume of 14 mL and thermostated at 20 ◦C for approximately 2 h. Two
independent process runs were carried out where 15 and 14 off-
line samples were taken during the first and during the second run,
respectively. 40 mM/20 mM benzaldehyde and 400 mM acetalde-
hyde were prepared in 50 mM phosphate buffer pH 7.5 containing
2 mM MgSO4 and 0.5 mM ThDP. The reaction was started by the
addition of wtBFD (8 U) dissolved in the same buffer; 50 mM phos-
phate buffer pH 7.5 containing 0.5 mM ThDP and 2 mM MgSO4.
Here, one unit of activity is defined as the amount of enzyme which
catalyzes the formation of 1 ␮mol of 2-hydroxypropiophenone
(HPP) in 1 min at 30 ◦C. Activity analysis for 2-HPP synthesis was
performed with 40 mM benzaldehyde and 400 mM acetaldehyde in
50 mM phosphate buffer containing 0.5 mM ThDP and 2 mM MgSO4
at pH 7.5.
2.2.3. Off-line analysis
Reactions were analyzed by HPLC (Agilent 1100, Hewlett
Packard) equipped with a LiChrosphere RP-8 column (Hypersil,
250 mm × 4 mm, Merck) using triethanolamine (0.2%, pH 3.5):
methanol (50:50, v/v) as an eluent at a flow rate of 1.0 mL min−1 at
40 ◦C. Typical retention times were: 2-HPP 4.9 0.1 min, benzalde-
hyde 5.7 0.1 min. The reactions were quenched by the addition
of a stop-buffer (90% acetonitrile, 5% H2O, and 5% H3PO4) with the
ratio of 1:1 (sample: stop-buffer, v/v) followed by intense mixing
and centrifugation of the precipitate. During the process runs 15
(first process) or 14 (second process) samples were taken simul-
taneously to one of the 63 fluorescence spectra or 48 fluorescence
spectra, respectively.
Limited work has been published on the use of spectroscopic
methods in biocatalysis up to now. Due to disturbance of water
IR-spectroscopy gives often suboptimal results. However, this com-
munication demonstrates the successful monitoring of the 2-HPP
synthesis catalyzed by wtBFD via 2D-fluorescence spectroscopy
where reactions run in aqueous medium.
2. Experimental
2.1. Material
All chemicals were purchased from Sigma–Aldrich (Schnelldorf,
Germany) and Carl Roth GmbH (Karlsruhe, Germany).
2.2.4. On-line monitoring via fluorescence spectroscopy and
chemometry
2.2. General methods
On-line monitoring of the carboligations catalyzed by wtBFD
was achieved using a 2D-fluorescence sensor (BioView® sensor).
The fluorescence sensor is equipped with a xenon flash lamp for
the excitation light. The sensor is connected directly by a bifurcated
liquid light conductor (Lumatec, Germany) to a quartz window in a
17-mm electrode standard port of a steel reactor. The sensor uses
two independently rotating filter wheels with 16 different filters
for excitation and emission and a photomultiplier for detection of
emission light. Each spectrum has 150 combinations of excitation
(Ex) emission (Em) wavelength pairs in a region of 270–550 nm
for excitation and in the region of 310–590 nm for emission with
a bandwidth of 20 nm. Intensities obtained from full excitation
and emission wavelength range are also included. Full fluorescence
spectra were recorded every 1.5 min during biotransformations.
Thus, 63 spectra of the first process and 48 spectra of the second
process were collected. To eliminate the differences in fluorescence
intensities occurred by varying the initial conditions, difference
spectra to the first spectra of each process are calculated.
2.2.1. Overexpression, fermentation and purification
Plasmid hosting wild type BFD was kindly supplied by PD. Dr.
Martina Pohl, Institute of Molecular Enzyme Technology/Heinrich-
Heine University of Düsseldorf, Germany. Precultivation of the
variants was done in shaking cultures (200 mL, pH 7.0 in 1 L
flasks) using a complex batch medium containing glucose (2.0 g/L),
MgSO4·7H2O (1.0 g/L), NH4Cl (0.2 g/L), (NH4)2SO4 (2.0 g/L), KH2PO4
(13 g/L), K2HPO4 (10 g/L), NaH2PO4·H2O (6 g/L), yeast extract
(3 g/L), ampicillin (0.1 g/L), and kanamycin (0.05 g/L). Fed-batch fer-
mentation was performed in a 2 L foil-fermenter (Bioengineering
AG, Wald, Switzerland) using the complex medium defined before
with 0.2% (v/v) Desmophen as an antifoam agent and 25% ammonia
solution for pH control at 7.1. The fermenter was inoculated by the
addition of 1% (v/v) active preculture having an OD600 nm = 1.0. After
12 h the feed solution was supplied containing glucose (600 g/L),
yeast extract (15 g/L), MgSO4·7H2O (400 g/L), ampicillin (0.1 g/L),

126
S. Kara et al. / Journal of Molecular Catalysis B: Enzymatic 66 (2010) 124–129
Fig. 3. Course of fluorescence intensities in the regions of Ex390 nm/Em450 nm,
Ex370 nm/Em450 nm and Ex370 nm/Em430 nm during the carboligation of ben-
zaldehyde and acetaldehyde (first process) catalyzed by wtBFD.
Fig. 2. General procedure for the development of the chemometric model in order
to predict the concentrations (dashed line).
intensity (RFI) units was more significant, remarkable increases in
RFI units were also detected in the regions of Ex370 nm/Em430 nm
and Ex370 nm/Em450 nm. Fig. 3 displays the increase in RFI units in
the described regions from the first process run. In Fig. 3 next to the
general increase of the fluorescence intensities, noise in spectra can
be seen. In order to avoid loss of information related to other Ex/Em
regions all collected wavelength pairs were used for the develop-
ment of a chemometric model since chemometric modeling can
handle noisy spectra very well.
Background spectra of reaction buffer (50 mM phosphate buffer
with cofactors ThDP and Mg+2), substrate solutions (benzaldehyde
or 3-methoxybenzaldehyde with acetaldehyde), as well as 2-HPP
product solution were analyzed. The remarkable intensity value
of the buffer system (2730 RFI unit; Ex390 nm/Em450 nm) can be
explained due the presence of cofactor ThDP which is the biologi-
cally active form of vitamin B1 (given in supporting information).
Principle component analysis (PCA) and partial least square
(PLS) regression are widely used techniques in multivariate data
analysis in order to reduce the huge amount of data to a small num-
ber of components [12]. The analysis of the fluorescence spectra
was achieved by PCA and PLS regression. To evaluate the number of
latent variables (principal components) and to compare both pro-
cesses a principle component analysis is performed. Chemometric
models, like PLS are developed to correlate the information from
the fluorescence spectra with the off-line data set of the respective
process variable. In the present work two process runs were ana-
lyzed where off-line samples were used to build a well performance
model. For each process variable (substrate and product) an own
PLS regression model was performed using the data of both process
runs together. The model was developed using multilinear partial
least square regression in the MathWorksTM MATLAB® utilizing the
N-way toolbox [13] (Fig. 2) with unfolded spectra.
3. Results and discussion
3.2. Principal component analysis (PCA) of fluorescence data
Fluorescence spectroscopy has been applied so far to monitor
fluorophores such as amino acids (e.g., tryptophan, tyrosine, and
phenylalanine), vitamins (e.g., pyridoxine (vitamin B6), riboflavin
(vitamin B2)) and coenzymes (e.g., NAD(P)H, FAD, and FMN). Main
applications of fluorescence spectroscopy are found for monitor-
ing of the cultivations where cell dry mass concentration, product
concentrations, and the actual metabolic state of the cell are fol-
lowed. In this contribution the applicability of 2D-fluorescence
spectroscopy to monitor the synthesis of 2-hydroxyketone was
demonstrated without addition of a fluorescence marker to follow
the reaction components.
For evaluation of the fluorescence data and for comparison of
both processes principle component analysis is performed. In the
score plot of the raw data both process runs are separated by the
second principal component which indicates that both processes
have different initial conditions (Fig. 4, left). Thus, the difference
in starting benzaldehyde concentrations is recognized which is
the only varying parameter between two process runs. However,
small deviations in reaction temperatures (20.3 ◦C, first process and
19.9 ◦C, second process) have also been taken into account since
fluorescence measurements are sensitive to any temperature devi-
ation. Different from the PCA based on the raw data, when PCA is
performed based on the difference spectra both process runs show
similarities with some runaway spectra. The score plot of the differ-
ence spectra demonstrates that a clear separation is not observed
(Fig. 4, right). In order to predict the latent information PCA based
on the raw data shows that three components are required whereas
PCA based on the difference spectra indicates that two components
are adequate.
The difference between two processes can be displayed as a
difference spectrum between the respective initial spectra of two
reactions runs (Fig. 5). In most Ex/Em regions lower RFI units were
observed which can be resulted from the small deviation in reac-
tion temperatures. However, significant differences were observed
in the regions of Ex370 nm/Em450 nm and Ex450 nm/Em530 nm
where these regions are well known for NADH and flavin, respec-
tively.
3.1. Monitoring of carboligation reactions catalyzed by wtBFD
The best of our knowledge, the monitoring of the carboligation
reactions via 2D-fluorescence spectroscopy yielding asymmetric
hydroxyketones has been achieved for the first time. The synthesis
of 2-HPP which is the main carboligation product starting from ben-
zaldehyde and acetaldehyde was successfully followed by on-line
analysis.
During the wtBFD catalyzed asymmetric C–C bond formations
2D-fluorescence measurements were performed. A steady increase
in fluorescence intensities in the region of Ex390 nm/Em450 nm
was detected during the course of both biotransformations
and verified by highest PCA-loadings. Besides the region of
Ex390 nm/Em450 nm where the increase in relative fluorescence

S. Kara et al. / Journal of Molecular Catalysis B: Enzymatic 66 (2010) 124–129
127
Fig. 4. PCA analysis of the fluorescence spectra. The score plot of the raw data (left) and the score plot of the difference spectra (right).
Table 1
RMSEC of process variables of carboligations starting with ben-
zaldehyde and acetaldehyde catalyzed by wtBFD.
Process variable
RMSEC
2-HPP
Benzaldehyde
1.080 mM (4.73%)
0.765 mM (2.25%)
centrations of the reaction components could be predicted on the
basis of the collected fluorescence spectra. Subsequently, two PLS
regression models were developed to correlate the measured con-
centrations of the carboligation product 2-HPP and the substrate
benzaldehyde with the courses of fluorescence intensities in differ-
ence spectra. Two latent variables were adequate to obtain a good
correlation. Fig. 6 demonstrates the good calibration of off-line data
and on-line spectra for the synthesis of 2-HPP and conversion of
benzaldehyde for the first process run.
Similarly, Fig. 7 shows the calibration of 2-HPP and the substrate
benzaldehyde for the second process.
The quality of prediction was quantified by analyzing the root
mean square error of calibration (RMSEC) determined from the
square root of the mean squared deviations between the off-line
measurements and the on-line determinations (Table 1).
The internal validation (full-cross validation) of the developed
chemometric model exhibits the quality of the predicted concen-
trations for the time intervals where no off-line information was
available. Thus, internal validation reveals the model performance
over the whole examined time range. Besides the internal valida-
tion, an external validation of the chemometric model was analyzed
which allows us to investigate the quality of the developed model
to be applied for the next set of experiments. Thus, using the on-line
fluorescence data of the next reaction with the previously devel-
oped model the concentrations of the reaction components of the
next reaction were predicted. Developed models were assured by
Fig. 5. Difference spectrum of two process runs displayed with respect to their initial
spectra.
Since the analyzed reactions are only differ in initial substrate
concentrations, observed high intensities in the difference spec-
trum can only result from benzaldehyde. Therefore, for validation
of 2-HPP concentrations a chemometric model is developed based
on the difference spectra whereas for validation of benzaldehyde
concentrations a model is built based on the raw data.
3.3. Chemometric modeling for the fluorescence based on-line
monitoring
Fluorescence spectroscopy offers the possibility to perform
multi-wavelength monitoring of the bioprocesses. Using suitable
mathematical tools such as chemometric regression models con-
Fig. 6. Analysis of 2-hydroxypropiophenone synthesis (left) and benzaldehyde conversion (right) for the first process. On-line predicted data and off-line HPLC data.

128
S. Kara et al. / Journal of Molecular Catalysis B: Enzymatic 66 (2010) 124–129
Fig. 7. Analysis of 2-hydroxypropiophenone synthesis (left) and benzaldehyde conversion (right) for the second process. On-line predicted data and off-line HPLC data.
Fig. 8. The carboligation of 3-methoxybenzaldehyde and acetaldehyde catalyzed by wtBFD.
Fig. 9. Analysis of 3-methoxybenzaldehyde conversion for the synthesis of 2-hydroxy-1-(3-methoxyphenyl)propan-1-one, first process (left) and second process (right).
On-line predicted data and off-line HPLC data.
performing full-cross validation where observed noises were elim-
inated reducing the quality of prediction.
was not available for calibration, Fig. 9 demonstrates only the
off-line determined concentrations and the on-line predicted
data of 3-methoxybenzaldehyde. For 3-methoxybenzaldehyde an
own PLS regression model was developed for each process.
Self-carboligation of 3-methoxybenzaldehyde yielding 2-hydroxy-
1,2-bis(3-methoxyphenyl)-ethanone was not detected in any reac-
tion analyzed (chromatograms are found in provided supporting
information).
RMSEC for the prediction of 3-methoxybenzaldehyde conver-
sion was determined as 0.199 mM (2.00%) for the first process and
0.169 mM (1.05%) for the second process which demonstrates that
the applicability of fluorescence spectroscopy can be broadened to
substituted benzaldehyde derivatives.
A wide variety of benzaldehyde derivatives are accepted by
wtBFD [9(a)]. The highest conversions were detected with meta-
substituted benzaldehyde derivatives whereas ortho-substituted
benzaldehyde derivatives were mainly identified as bad substrates
[9(a)]. The influence of electronic effects on the enantiomeric
excess was demonstrated to depend on the Hammett-correlation
as well as on steric effects of substituents on the aromatic sub-
strates [9(a,b)]. Thus, alternative to benzaldehyde, carboligation of
3-methoxybenzaldehdye (m-anisaldehyde) and acetaldehyde was
analyzed (Fig. 8). Addition of any organic solvent was not required
since 3-methoxybenzaldehyde was soluble at analyzed reaction
conditions.
Two reactions were carried out containing 10 mM/16 mM 3-
methoxybenzaldehyde and 400 mM acetaldehyde prepared in
50 mM phosphate buffer in the presence of the cofactors 0.5 mM
ThDP, 2 mM MgSO4 at pH 7.5 and 20 ◦C. During analyses 16 off-
line samples were taken with previously described procedure
until 90 min. The biotransformation was started by the addi-
tion of wtBFD (15 U) solution dissolved in 50 mM phosphate
buffer, 0.5 mM ThDP and 2 mM MgSO4 at pH 7.5. Since the
reference 2-hydroxy-1-(3-methoxyphenyl)propan-1-one product
4. Conclusions
On-line analysis of bioprocesses is of high interest that enables
direct control of process parameters in order to achieve improved
process efficiencies. Among the developed on-line analysis tech-
niques fluorescence spectroscopy has been mainly applied in
monitoring of cultivations to obtain high biomass yield. The biggest
advantage of fluorescence spectroscopy is its applicability to aque-
ous systems providing high quality spectra of all fluorescent

S. Kara et al. / Journal of Molecular Catalysis B: Enzymatic 66 (2010) 124–129
129
compounds at low concentrations. In this contribution the appli-
cability of fluorescence spectroscopy for monitoring the synthesis
of 2-hydroxypropiophenone (HPP), a chiral 2-hydroxyketone, was
successfully demonstrated. Wild type BFD catalyzed carboliga-
tions were monitored by the BioView® fluorescence sensor (DELTA,
Denmark) with a very good correlation. A reliable chemometric
model was developed where the concentrations of reaction com-
ponents were validated with satisfying RMSEC values. Besides the
synthesis of 2-HPPs which is the main carboligation product, the
conversion of benzaldehyde was also successfully monitored via
2D-fluorescence spectroscopy. Therefore, this study demonstrates
for the first time that 2D-fluorescence spectroscopy can be suc-
cessfully applied in biocatalyzed asymmetric C–C bond formations,
even in water as reaction medium with low starting material con-
centrations. Via chemometric models concentrations/conversions
of reaction components can be predicted in a small range of errors.
On-line determined data for substrate conversion and product for-
mation during the course of biotransformations can be used in
progressive curve analyses. Thus, the enzyme kinetic parameters
can be further identified by developing appropriate kinetic models
which is going to be included in our next communication.
(b) T. Knüttel, H. Meyer, T. Scheper, Anal. Chem. 77 (2005) 6184–6189;
(c) F.R.P. Oliveira, K. Goldberg, A. Liese, B. Hitzmann, Chemometr. Intell. Lab.
Syst. 94 (2) (2008) 118–122;
(d) A. Alonso-Mateos, M.J. Almendral-Parra, Y. Curto-Serrano, F.J. Rodríguez-
Martín, J. Fluoresc. 18 (2008) 183–192;
(e) O. Divya, A.K. Mishra, Appl. Spectrosc. 62 (2008) 753–758.
[3] M. del Busto-Ramos, M. Budzik, C. Corvalan, M. Morgan, R. Turco, D. Nivens, B.
Applegate, Appl. Microbiol. Biotechnol. 78 (2008) 573–580.
ˇ
ˇ
[4] V. Stefuca, I. Vosˇtiar, J. Sefcˇovicˇová, J. Katrlík, V. Mastihuba, M. Greifová, P.
Gemeiner, Appl. Microbiol. Biotechnol. 72 (2006) 1170–1175.
[5] K. Kivijarju, K. Salonen, U. Moilanen, T. Eerikäinen, J. Ind. Microbiol. Biotechnol.
35 (2008) 657–665.
[6] (a) J.R. Workman, J. Jerome, Appl. Spectrosc. Rev. 34 (1) (1999) 1–89;
(b) M.R. Dadda, D.C.A. Sharp, A.J. Pettman, C.J. Knowles, J. Microbiol. Methods
41 (2000) 69–75;
(c) L. Mariey, J.P. Signolle, C. Amiel, J. Travert, Vib. Spectrosc. 26 (2001) 151–159.
[7] (a) S. Marose, C. Lindemann, T. Scheper, Biotechnol. Prog. 14 (1) (1998) 63–74;
(b) T. Scheper, B. Hitzmann, E. Stärk, R. Ulber, R. Faurie, P. Sosnitza, K.F. Reardon,
Anal. Chim. Acta 400 (1999) 121–134;
(c) E. Stärk, B. Hitzmann, K. Schügerl, T. Scheper, C. Fuchs, D. Köster, H. Märkl,
Adv. Biochem. Eng./Biotechnol. 74 (2002) 21–38;
(d) K. Hantelmann, M. Kollecker, D. Hüll, B. Hitzmann, T. Scheper, J. Biotechnol.
121 (2006) 410–417;
(e) M.B. Haack, A. Eliasson Lantz, P.P. Mortensen, L. Olsson, Biotechnol. Bioeng.
96 (5) (2007) 904–913.
[8] (a) A. Liese, K. Seelbach, C. Wandrey, Industrial Biotransformations, 2nd edi-
tion, Wiley-VCH, 2006, pp. 447–448;
(b) T. Stillger, Enantioselektive C–C Knüpfung mit Enzymen, Dissertation, For-
shungszentrum Jülich GmBH, 2006, ISBN: 3-89336-457-459.
[9] (a) H. Iding, T. Dünnwald, L. Greiner, A. Liese, M. Müller, P. Siegert, J. Grötzinger,
A.S. Demir, M. Pohl, Chem. Eur. J. 6 (8) (2000) 1483–1495;
(b) A. Liese, Biological Principles Applied to Technical Asymmetric Cataly-
sis, Verlag Forschungszentrum Jülich, 2003, 207 S., ISBN: 3-89336-344-0, pp.
49–61;
Acknowledgements
We thank Prof. Dr. Martina Pohl (Institute of Molecular Enzyme
Technology, Heinrich-Heine University of Düsseldorf, Research
Center Jülich) for kindly providing the wild type BFD.
(c) B. Lingen, M. Pohl, A.S. Demir, A. Liese, M. Müller, Oxidative Stress Dis. 11
(2004) 113–129 (Thiamine);
(d) P. Domínguez de María, T. Stillger, M. Pohl, M. Kiesel, H. Gröger, H. Trauth-
wein, Adv. Synth. Catal. 250 (2008) 165–173.
[10] N. Kurlemann, A. Liese, Tetrahedron: Asymmetry 15 (2004) 2955–2958.
[11] M.M. Bradford, Anal. Biochem. 72 (1976) 248–254.
[12] (a) S. Wold, M. Sjostrom, L. Eriksson, Chemometr. Intell. Lab. Syst. 58 (2) (2001)
109–130;
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.molcatb.2010.04.006.
(b) D. Solle, D. Geissler, E. Stärk, T. Scheper, B. Hitzmann, Bioinformatics 19 (2)
(2003) 173–177;
(c) D. Boehl, D. Solle, B. Hitzmann, T. Scheper, J. Biotechnol. 105 (1–2) (2003)
179–188;
References
(d) J.I.I. Rhee, T.-H. Kang, Process Biochem. 42 (7) (2007) 1124–1134.
[13] (a) C.A. Andersson, R. Bro, Chemometr. Intell. Lab. Syst. 52 (2000) 1–4;
(b) D. Baunsgaard, C.A. Andersson, A. Arndal, L. Munck, Food Chem. 70 (1) (2000)
113–121.
[1] A. Kandelbauer, W. Kessler, R.W. Kessler, Anal. Bioanal. Chem. 390 (2008)
1303–1315.
[2] (a) T. Knüttel, T. Hartmann, H. Meyer, H.T. Scheper, Enzyme Microb. Technol.
29 (2001) 150–159;