Synthesis of a cationic pyridoxamine conjugation reagent and application to the mechanistic analysis of an artificial transaminase

Article abstract of DOI:10.1016/S0960-894X(00)00419-4

An N-methylated, cationic pyridoxamine conjugation reagent was synthesized and tethered via a disulfide bond to a cysteine residue inside the cavity of intestinal fatty acid binding protein. The conjugate was characterized and the kinetic parameters compared to its nonmethylated pyridoxamine analogue. Kinetic isotope effects were used for further mechanistic analysis. Taken together, these experiments suggest that a step distinct from deprotonation of the ketimine in the pyridoxamine to pyridoxal reaction is what limits the rate of the artificial transaminase IFABP-Px. However, the internal energetics of reactions catalyzed by the conjugate containing the N-methylated cofactor appear to be different suggesting that the MPx reagent will be useful in future experiments designed to alter the catalytic properties of semisynthetic transaminases. (C) 2000 Elsevier Science Ltd.

Full text of DOI:10.1016/S0960-894X(00)00419-4

Bioorganic & Medicinal Chemistry Letters 10 (2000) 2091±2095
Synthesis of a Cationic Pyridoxamine Conjugation Reagent
and Application to the Mechanistic Analysis of an
Arti®cial Transaminase
Hao Kuang, Dietmar Haring, Dongfeng Qi, Aram Mazhary and Mark D. Distefano*
Department of Chemistry, University of Minnesota, 207 Pleasant Street SE, Minneapolis, MN 55455, USA
Received 28 April 2000; accepted 11 July 2000
AbstractÐAn N-methylated, cationic pyridoxamine conjugation reagent was synthesized and tethered via a disul®de bond to a
cysteine residue inside the cavity of intestinal fatty acid binding protein. The conjugate was characterized and the kinetic parameters
compared to its nonmethylated pyridoxamine analogue. Kinetic isotope e€ects were used for further mechanistic analysis. Taken
together, these experiments suggest that a step distinct from deprotonation of the ketimine in the pyridoxamine to pyridoxal reaction
is what limits the rate of the arti®cial transaminase IFABP-Px. However, the internal energetics of reactions catalyzed by the con-
jugate containing the N-methylated cofactor appear to be di€erent suggesting that the MPx reagent will be useful in future experi-
ments designed to alter the catalytic properties of semisynthetic transaminases. # 2000 Elsevier Science Ltd. All rights reserved.
Although native enzymes are versatile and highly e-
cient catalysts, they are also limited in terms of the
reaction types they catalyze and the substrate selectivity
that they manifest. In order to design tailor-made bio-
catalysts, a variety of methods have been developed
including RNA aptamers, catalytic antibodies and
cyclodextrin-based enzyme mimics; the chemical intro-
duction of a synthetic cofactor into an existing protein
has proven to be particularly versatile. Such `semisyn-
thetic enzymes' use the framework of well-known pro-
teins for binding the substrate in a highly chiral
environment and creating an arti®cial active site.1 3
Recently, we have expanded this concept by combining
chemical and genetic enzyme engineering: a cysteine
residue can be introduced by site-directed mutagenesis
into a speci®c position in a protein and used as point of
attachment for the chemical tethering of a desired catalytic
active group.4
by site-directed mutagenesis and linked to a pyridox-
amine-moiety (Px) via a disul®de bond (Fig. 1).4 The
resulting arti®cial enzymes (IFABP-Px) catalyzed the
transamination of a-keto and amino acids similar to
native aminotransferases. The reactions were most e-
cient when the pyridoxamine was linked to Cys 60
which is located in the middle of the ®rst b-sheet ele-
ment that forms the cavity.5 Here, we describe the
synthesis of a modi®ed arti®cial pyridoxamine cofactor,
its conjugation to IFABP and the implications on the
catalytic mechanism of the resulting arti®cial transami-
nase. These experiments highlight the versatility of the
`semisynthesis' approach which allows the use of both
chemical synthesis and genetic engineering to prepare
catalytic systems.
The mechanism of the pyridoxamine-catalyzed transa-
mination of a-keto and amino acids consists of two
reversible half reactions (Scheme 1). Early studies of
pyridoxal derivatives have shown, that pyridine-4-alde-
hyde does not catalyze the reaction, whereas 1-methyl-
4-formylpyridinium iodide is able to form an aldimine
similar to 4 with amino acids, which is converted in a
second, slower reaction to a ketimine corresponding
with 2.6 This di€erence was attributed to the greater
electron withdrawing ability of the quaternized pyridine
nitrogen. Comparing the transamination reactions of 3-
hydroxy-4-formylpyridine with 1-methyl-3-hydroxy-4-
formylpyridinium chloride, the initial rates of reactions
promoted by the N-methylated species were up to 20-
The rat intestinal fatty acid binding protein (IFABP)
consists of two orthogonal b-sheets which form a large
cavity of approximately 600 A3. The entrance to the
protein interior is covered by a ¯exible a-helical loop, so
that the bound substrates (e.g., fatty acids) are com-
pletely sequestered within. In earlier work, a cysteine
residue was placed in various positions inside the cavity
*Corresponding author. Tel.: +1-612-624-0544; fax: +1-612-626-
7541; e-mail: distefan@chem.umn.edu
0960-894X/00/$ - see front matter # 2000 Elsevier Science Ltd. All rights reserved.
PII: S0960-894X(00)00419-4

2092
H. Kuang et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2091±2095
that the overall folding of these chemically modi®ed
proteins resembles the native state.13
To assess the in¯uence of the cationic pyridoxamine
moiety inside IFABP on the protein stability, the prop-
erties of the unfolding transition were examined. Mea-
suring tryptophan ¯uorescence, the guanidine hydro-
chloride (GdnHCl) induced unfolding of several IFABP
mutants including V60C was recently studied.14 In those
experiments, the ¯uorescence data were ®tted to a two
state model which suggests that only the native and the
fully unfolded forms of the protein are present in sig-
ni®cant concentrations at equilibrium.15 A transition
midpoint concentration of 1.18 M GdnHCl was found
for IFABP-V60C, indicating a small decrease in protein
stability compared to the wild-type IFABP (1.36 M).
Using the same experimental conditions, we determined
the midpoint concentrations of IFABP-Px and-MPx to
be 1.13Æ0.14 and 1.22Æ0.13 M GdnHCl, respectively
(Fig. 3). These data demonstrate that neither the
pyridoxamine group nor the cationic nitrogen tethered
Figure 1. The intestinal fatty acid binding protein mutant Val60Cys
(IFABP-SH) was chemically tethered to a pyridoxamine cofactor (Px)
forming the arti®cial transaminase IFABP-Px.
fold increased depending on the pH.7 However, when
the pyridoxal phosphate cofactor present in aspartate
aminotransferase was substituted by its N-methylated
analogue, the enzyme lost >99.8% of its catalytic
activity.8,9 Thus N-quaternization may have both posi-
tive and negative in¯uence on transamination reactions.
To determine the e€ect of N-methylation on the cataly-
tic properties of IFABP-Px we synthesized a suitable
conjugation reagent, 1-methyl-5-(2-pyridyldithio)pyr-
idoxamine 9 (Scheme 2). The triacetylated intermediate
6 was synthesized from the commercially available pyr-
idoxamine dihydrochloride in four steps as described
previously.10 N-Methylation was achieved using methyl
iodide in benzene11 followed by acid hydrolysis of the
acetyl groups. Finally, the free thiol of 8 was activated
for conjugation reactions by conversion to the disul®de
9.12 The cationic conjugation reagent was puri®ed by
preparative reversed-phase HPLC and an overall yield
of 19% (based on 6) was obtained.
The IFABP-mutant V60C was expressed in E. coli cul-
tures and puri®ed to homogenity.5 After incubating the
conjugation reagent 9 (MPx) with the protein for 18 h at
room temperature, a disul®de bond with the single
cysteine residue in position 60 inside the cavity of IFABP
was formed.10 The excess of 9 was removed from the
IFABP-MPx conjugate by gel ®ltration chromato-
graphy. The conjugation eciency was 92% as deter-
mined by titration of the free thiol with 5,50-dithiobis(2-
nitrobenzoic acid) before and after the conjugation
reaction.10 The UV spectrum showed the absorption
bands of both the protein (280 nm) and the bound MPx
(338 nm), the latter one being red-shifted compared to the
nonmethylated IFABP-Px (Fig. 2). A positive band in the
CD spectra of IFABP-Px and-MPx indicates the posi-
tioning of the cofactors within a chiral environment, i.e.,
the protein cavity (Fig. 2). Finally, the elution volumes
obtained by gel permeation chromatography of IFABP-
V60C and both conjugates were identical, suggesting
Scheme 2. Synthesis of the cationic pyridoxamine conjugation reagent
reagent 9 (MPx).
Figure 2. UV and CD spectra of IFABP-MPx (solid) and IFABP-Px
(dashed). The spectra were taken in 20 mM HEPES pH 7.5 with 50 mM
(UV) or 20 mM (CD) conjugate.
Scheme 1. Reaction mechanism of the pyridoxamine-catalyzed transamination.

H. Kuang et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2091±2095
2093
Figure 3. Guanidinium hydrochloride-induced equilibrium unfolding
of IFABP-MPx monitored by tryptophan ¯uorescence (Ex. 280 nm,
Em. 338 nm). Samples contained protein (1.0 mM), Tris HCl (50 mM,
pH 7.4), EDTA (0.1 mM) and guanidinium chloride (0±5.0 M). Data
were ®tted to a two state model, previously used for lipid binding
proteins.14
Figure 4. Comparison of glutamic acid production in the catalytic
reaction performed by IFABP±MPx (A) or IFABP-Px (B) using tyr-
osine as amino acid. After 120 h, 16.5 turnovers (70% ee) for the MPx-
conjugate and 16 turnovers (92% ee) for the Px-conjugate were
achieved. Reactions were performed with 50 mM catalyst, 50 mM a-
keto glutarate and 5 mM tyrosine in 0.2 M HEPES pH 7.71 at 37 ꢀC.
.
inside the hydrophobic protein cavity have a destabiliz-
ing e€ect on the protein.
methylation had no positive e€ects on the reaction rate.
This suggests that the deprotonation of 2 or 4 are not
the rate limiting steps in the IFABP-Px catalyzed trans-
aminations. To examine the possibility that proton
transfer reactions are rate determining in the reaction
catalyzed by IFABP-Px, kinetic isotope e€ect experi-
ments (KIE) were undertaken (Table 2).
When IFABP-MPx was incubated at 37 ꢀC with a-
ketoglutarate and phenylalanine or tyrosine, respec-
tively, catalytic transamination was observed. Table 1
lists the number of turnovers found and the enantio-
meric excess of the l-glutamate produced in 24 h.
Inspection of the values given in Table 1 indicates that
the overall transamination rates exhibited by IFABP-Px
and IFABP-MPx are similar while the enantioselectivity
of the latter is signi®cantly lower. Although the above
results were obtained after 24 h of incubation, it should
be noted that these constructs are active for at least ®ve
days under these reaction conditions; after 120 h approxi-
mately 16 turnovers occur for each catalyst (Fig. 4). To
examine the kinetics of these reactions in more detail,
the initial rates were determined by analyzing the amount
of glutamate formed after 6h of reaction (Table 1). After
this short time less than one turnover is observed and
thus these kinetic parameters re¯ect events occuring
only in the ®rst half reaction (cf. Scheme 1 from left to
right). In comparison to the IFABP-Px, the N-methy-
lated conjugate showed higher values (1.1- to 3.8-fold)
First, it should be noted that the transamination of a-
ketoglutarate and phenylalanine to glutamic acid and
phenylpyruvate is essentially linear for 100 h at which
point at least 10 turnovers have occured. If the second
half reaction (pyridoxal to pyridoxamine) was rate
determining, the reaction rate would decrease after one
turnover. Thus it appears that the pyridoxal to pyr-
idoxamine reaction is not rate determining using these
substrates. However, in reactions using a-ketoglutarate
and alanine as substrates, the reaction rate is clearly
biphasic with a fast phase in the ®rst turnover followed
by a slower phase in subsequent turnovers. These latter
results illustrate that the pyridoxal to pyridoxamine
reaction can be rendered to be rate determining by using
a poor substrate.16 KIE measurements using [2H8]-phe-
nylalanine give a VH/VD of unity. Since an isotope e€ect
would only be observed with this substrate if the pyr-
idoxal to pyridoxamine reaction was rate determining,
the absence of a KIE is consistant with the above
assertion that the pyridoxamine to pyridoxal reaction is
rate determining when a-ketoglutarate and phenylala-
nine are used as substrates. Interestingly, experiments
using [2-2H]-alanine showed no signi®cant KIE, sug-
gesting that although the pyridoxal to pyridoxamine
reaction may be rate determining using a-ketoglutarate
0
0
for Km , whereas the kcat values were similar for both
catalysts.
The greater electron withdrawal of the quaternized pyr-
idine nitrogen was expected to facilitate the deprotona-
tion of ketimine 2 in the benzylic position (2!3) as well
as the deprotonation of aldimine 4 at the Ca of the
amino acid (4!3). However, the kinetic data for the
pyridoxamine to pyridoxal half reaction showed that N-
Table 1. Kinetic parameters of the transamination reactions catalyzed by IFABP-Px and-MPxa
0
0
1
0
1
1
Turnover/eeb
Km mM
kcat
h
kcat0/Km
h
mM
IFABP-Px
Phe
Tyr
Phe
Tyr
3.9/93%
4.2/93%
5.8/41%
1.7/60%
1.8Æ0.5
4.2Æ0.9
6.8Æ1.3
4.5Æ1.3
0.29Æ0.02
0.26Æ0.02
0.23Æ0.01
0.17Æ0.01
0.16
0.06
0.03
0.04
IFABP-MPx
aThe reactions were performed with 50 mM IFABP conjugate, varying concentrations of a-ketoglutarate and 5 mM amino acid in 0.2 M HEPES pH
7.5 at 37 ꢀC (100 mL total volume). The formation of glutarate was monitored by HPLC.10 The reactions showed saturation kinetics and the data
points were ®tted to a standard Michaelis±Menten equation.
bThe reactions were stopped after 24 h and the production of glutarate analyzed by HPLC. The enantiomeric excess is that of l-glutarate.

2094
H. Kuang et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2091±2095
Table 2. Primary kintic isotope e€ects of the transamination reaction catalyzed by IFABP-Pxa
Substrate/Solvent
IFABP-Px
IFABP-MPx
Rate mM/h
Rate mM/h
VH/VD
VH/VD
Phenylalanine
[2H8]-Phenylalanine
Alanine
13.1Æ1.5
12.8Æ1.1
5.3Æ0.5
1.5Æ0.2
4.3Æ0.5
1.6Æ0.2
13.6Æ0.7
8.0Æ1.1
1.9Æ0.1
1.3Æ0.1
1.5Æ0.2
1.1Æ0.1
1.0Æ0.1
1.7Æ0.3
First phase
Second phase
First phase
[2-2H]-Alanine
1.2Æ0.2
1.3Æ0.2
1.2Æ0.1
Second phase
0.94Æ0.2
aThe reactions were performed with 50 mM IFABP-Px, 50 mM a-ketoglutarate and 5 mM amino acid in 0.2 M HEPES pH 7.5 at 37 ꢀC (100 mL total
volume). The formation of glutarate was monitored by HPLC.10
and alanine as substrates, the a-deprotonation step (that
is only one of several steps in the conversion between
pyridoxal and pyridoxamine) is not the rate determining
process in that reaction sequence. For comparison,
native aspartate aminotransferse from E. coli showed a
KIE of ꢁ2 with [2-2H]-aspartate as substrate, suggesting
that deprotonation of 4 to 3 is at least partially rate
determining.17,18
via a cysteine residue previously introduced by site-
directed mutagenesis. Although the cofactor contains a
permanent positive charge, the protein was not destabi-
lized. The N-methylated pyridoxamine moiety o€ered
no catalytic advantage in the transamination reactions
of the IFABP-MPx compared to the nonmethylated
analogue. This result is consistant with subsequent KIE
experiments that indicate that proton transfer steps are
not rate determining in the case of this particular arti®-
cial enzyme. However, given that proton transfer events
are frequently rate determining in natural transami-
nases,17,18 it is likely that the MPx reagent will be useful
in future experiments as the catalytic eciency of these
semisynthetic enzymes is increased through additional
cycles of design.
Reactions using a-ketoglutarate and phenylalanine were
performed in D2O to measure the solvent KIE. Since,
after several turnovers, deuterium will become incorpo-
rated in the methylene position of pyridoxamine, rate
measurements in D2O can provide information about
whether deprotonation of the ketimine in the pyridox-
amine to pyridoxal direction is rate determining. Ana-
lysis of the progress curve for the reaction in D2O
showed biphasic behavior, starting with a fast phase in
the ®rst 24±48 h followed by a slower phase in the sub-
sequent 48 h. Comparison of these rates with those
measured in H2O reveal no signi®cant KIE in the ®rst
phase (VH/VD=0.91) and a modest KIE (VH/VD=1.7)
in the slower phase; the small magnitude of this value
suggests that deprotonation of the ketimine is not sub-
stantially rate determining. Taken together, these KIE
experiments performed on IFABP-Px suggest that a
step distinct from deprotonation of the ketimine in the
pyridoxamine to pyridoxal reaction is what limits the
rate of the arti®cial transaminase IFABP-Px. Given
these results, it is perhaps not surprising that the pre-
paration of IFABP-MPx did not result in an increase in
catalytic eciency. However, several experiments with
IFABP-MPx suggest that the modi®ed MPx cofactor
may signi®cantly rearrange the energetics in these
transamaination reactions. KIE measurements using
[2H8]-phenylalanine with IFABP-MPx give a VH/VD of
1.7. This value is signi®cantly greater than that observed
with IFABP-Px and approaches the value observed with
aspartate aminotransferase suggesting that the modi®ed
cofactor may not be facilitating a-deprotonation of the
aldimine as it was designed to do. Moreover, the
decrease in enantioselectivity in the IFABP-MPx cata-
lyzed reaction suggests that the modi®ed cofactor is either
altering the conformation of the quinoid intermediate
(resulting in less facial selectivity) or lowering the barrier
for racemization (as might occur if the acidity of the
benzylic protons in the ketimine intermediate increased).
Acknowledgements
This work was supported by the National Science
Foundation (CHE-9807495) and a fellowship of the
Deutsche Akademische Austauschdienst (NATO-Pro-
gramm).
References and Notes
1. Kaiser, E. T.; Lawrence, D. S. Science 1984, 226, 505.
2. Distefano, M. D.; Kuang, H.; Qi, D.; Mazhary, A. Curr.
Opin. Struct. Biol. 1998, 8, 8459.
3. Haring, D.; Schreier, P. Naturwissenschaften 1999, 86, 307.
4. Kuang, H.; Davies, R. R.; Distefano, M. D. Bioorg. Med.
Chem. Lett. 1997, 7, 2055.
5. Kuang, H.; Distefano, M. D. J. Am. Chem. Soc. 1998, 120,
1072.
6. Maley, J. R.; Bruice, T. C. J. Am. Chem. Soc. 1968, 90,
2843.
7. Maley, J. R.; Bruice, T. C. Arch. Biochem. Biophys 1970,
136, 187.
8. Furbish, F. S.; Fonda, M. L.; Metzler, D. E. Biochemistry
1969, 8, 5169.
9. Fonda, M. L. J. Biol. Chem. 1971, 246, 2230.
10. Kuang, H.; Brown, M. L.; Davies, R. R.; Young, E. C.;
Distefano, M. D. J. Am. Chem. Soc. 1996, 118, 10702.
11. Triacetylpyridoxamine (100 mg) was dissolved in 20 mL
hot benzene and 1 mL methyl iodine added. After 24 h of
re¯ux, the crude product 7 was puri®ed with a Sep-Pak C18
column (42% yield). 1H NMR (300 MHz, D2O) d 7.50 (s, 1H),
4.26 (s, 2H), 4.03 (s, 2H), 3.89 (s, 3H), 2.36 (s, 3H), 2.23 (s,
3H), 1.84 (s, 3H).
In conclusion, a new cationic pyridoxamine conjugation
reagent MPx 9 was synthesized and linked to a protein

H. Kuang et al. / Bioorg. Med. Chem. Lett. 10 (2000) 2091±2095
2095
12. The N-methylated compound 7 was hydrolyzed in 40 mL
of 48% aq HBr under re¯ux for 5 h. Excess water and HBr
were removed in vacuo. The resulting thiol 8 is instable and
was immediately dissolved in 6 mL degassed methanol. 2,20-
Dipyridyl disul®de (Aldrithiol-2; 500 mg in 15 mL methanol)
was added in 60 min under a nitrogen atmosphere at room
temperature. The solvent was removed, the yellow solid
washed with chloroform and dissolved in 2 mL water:aceto-
nitrile (1:1). 1-Methyl-5-(2-pyridyltdithio)pyridoxamine 9 was
puri®ed by RP-HPLC on a C18 column (solvent A: water:
acetonitrile:TFA 94.8:5:0.2; solvent B: acetonitrile:TFA 99.8:
0.2) with an elution pro®le of 10 min isocratic 100% A, fol-
lowed by a 50 min linear gradient to 40% B. Fractions of 9
were collected at 33 min, the solvent removed in vacuo and the
®nal conjugation reagent stored at 80 ꢀC (45% yield). 1H
NMR (300 MHz, D2O) d 8.14 (d, 1H, J=5.1), 7.93 (s, 1H),
7.61 (dt, 1H, J=7.7, 1.3), 7.37 (d, 1H, J=8.1), 7.13 (dd, 1H,
13. Gel permeation chromatography was performed on a
FPLC system with a Superose 12 HR 30 column. Samples
(100 mL) were eluted with 25 mM HEPES pH 7.5 and 150 mM
NaCl at a ¯ow rate of 0.5 mL/min. The column was calibrated
with bovine lung aprotinin (6.5 kDa), horse heart cytochrome c
(12.4 kDa), bovine erythrocyte carbonic anhydrase (29.0 kDa)
and bovine serum albumin (66.0 kDa).
14. Jiang, N.; Frieden, C. Biochemistry 1993, 32, 1115.
15. Ropson, I. J.; Gordon, J. I.; Frieden, C. Biochemistry
1990, 29, 9591.
16. Some aspartate aminotransferase mutants revealed also
distinctively di€erent reaction rates in both half reactions. For
examples, cf. Yano, T.; Kuramitsu, S.; Tanase, S.; Morino, Y.;
Kagamiyama, H. Biochemistry 1992, 31, 5878. Ziak, M.;
Jager, J.; Malashkevich, V. N.; Gering, H.; Jaussi, R.; Janso-
nius, J. N.; Christen, P. Eur. J. Biochem. 1993, 211, 475.
17. Kuramitsu, S.; Hiromi, K.; Hayashi, H.; Morino, Y.;
Kagamiyama, H. Biochemistry 1990, 29, 5469.
J=7.5, 5.1), 4.28 (s, 2H), 4.10 (s, 2H), 3.89 (s, 3H), 2.30 (s,
3H). UV (20 mM HEPES pH 7.5): e338=6210 M 1 cm
1
.
18. Goldberg, J. M.; Kirsch, J. F. Biochemistry 1996, 35, 5280.