Kinetics and mechanism of thermal decomposition of kynurenines and biomolecular conjugates: Ramifications for the modification of mammalian eye lens proteins

Article abstract of DOI:10.1039/b903196k

Thermal degradation reactions of kynurenine (KN), 3-hydroxykynurenine (3OHKN), and several adducts of KN, to amino acids and reduced glutathione (GSH) have been studied at physiological temperature. These compounds are all implicated in age-related mammalian eye lens cataract formation at the molecular level. The main reaction pathway for both KN and 3OHKN is deamination viaβ-elimination to carboxyketoalkenes CKA and 3OHCKA. These reactions show a weak pH dependence below pH values of ～8, and a strong pH dependence above this value. The 3OHKN structure deaminates at a faster rate than KN. A mechanism for the deamination reaction is proposed, involving an aryl carbonyl enol/enolate ion, that is strongly supported by the structural, kinetic, and pH data. The degradation of Lys, His, Cys and GSH adducts of the CKA moieties was also studied. The Lys adduct was found to be relatively stable over 200 h at 37 °C, while significant degradation was observed for the other adducts. The results are discussed in terms of known post-translational modification reactions of the lens proteins and compared to incubation studies involving KN and related compounds in the presence of proteins.

Full text of DOI:10.1039/b903196k

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/obc | Organic & Biomolecular Chemistry
Kinetics and mechanism of thermal decomposition of kynurenines and
biomolecular conjugates: Ramifications for the modification of mammalian
eye lens proteins†
Lyudmila V. Kopylova,a Olga A. Snytnikova,a Elena I. Chernyak,b Sergey V. Morozov,b Malcolm D. E. Forbes*c
and Yuri P. Tsentalovich*a
Received 16th February 2009, Accepted 14th April 2009
First published as an Advance Article on the web 1st June 2009
DOI: 10.1039/b903196k
Thermal degradation reactions of kynurenine (KN), 3-hydroxykynurenine (3OHKN), and several
adducts of KN, to amino acids and reduced glutathione (GSH) have been studied at physiological
temperature. These compounds are all implicated in age-related mammalian eye lens cataract formation
at the molecular level. The main reaction pathway for both KN and 3OHKN is deamination via
b-elimination to carboxyketoalkenes CKA and 3OHCKA. These reactions show a weak pH
dependence below pH values of ~8, and a strong pH dependence above this value. The 3OHKN
structure deaminates at a faster rate than KN. A mechanism for the deamination reaction is proposed,
involving an aryl carbonyl enol/enolate ion, that is strongly supported by the structural, kinetic, and
pH data. The degradation of Lys, His, Cys and GSH adducts of the CKA moieties was also studied.
The Lys adduct was found to be relatively stable over 200 h at 37 ◦C, while significant degradation was
observed for the other adducts. The results are discussed in terms of known post-translational
modification reactions of the lens proteins and compared to incubation studies involving KN and
related compounds in the presence of proteins.
spontaneous reactions, resulting in the formation of “secondary”
Introduction
UV filters (such as AHBG, AHBDG, 3OHKG-GSH).9–14
The main functions of the mammalian eye lens is to focus
The transparency of the lens and its high refractive index
light upon the retina and to filter ultraviolet light. The
are due to the compact packing of fiber cells that constitute
retina is protected from UV damage by low-molecular-weight
the lens nucleus. These cells are protein rich: proteins comprise
compounds contained in the lens, which absorb UV light
more than 30% of the wet lens weight. Most of the lens proteins
in the region 300–400 nm. The UV-filter compounds are
are crystallins—these are water-soluble structural proteins whose
kynurenine (KN) and its derivatives, which originate from the
main function is to increase the refractive index, while not
obstructing light.15 Once the proteins are produced, they form the
lens nucleus and the lens grows throughout the human lifespan
by the addition of new cell layers over the pre-existing core. Since
there is no protein turnover in the lens nucleus, post-translational
modifications, caused by oxidative processes, accumulate with age,
resulting in coloration, aggregation, and lower solubility of old
lens proteins. Eventually, these processes lead to the development
of age-related nuclear (ARN) cataracts.15–21
amino acid tryptophan. Scheme 1 shows UV-filter compound
formation and transformations. The most abundant UV filters
in the human lens are 3-hydroxykynurenine O-b-D-glucoside
(3OHKG), 4-(2-amino-3-hydroxyphenyl)-4-oxobutanoic acid
O-b-D-glucoside (AHBG), and glutathionyl-3-hydroxykynurenine
O-b-D-glucoside (3OHKG-GSH), followed by kynurenine (KN)
and 3-hydroxykynurenine (3OHKN).1–7 The initial steps in
Scheme 1 are enzymatic reactions: the formation of N-formylky-
nurenine (not shown) from TrpH is followed by hydrolysis
to yield KN. Hydroxylation and glucosylation of the latter
produce 3OHKN and 3OHKG.1,8 The deamination reactions
of the “primary” UV filters (KN, 3OHKN and 3OHKG) to
carboxyketoalkenes (CKAs) are followed by reduction and
linkage to free thiols (glutathione and cysteine). These are
Truscott and others have suggested that the UV-filter com-
pounds shown in Scheme
1 are important in crystallin
modification.22–25 Three of the major UV filters—KN, 3OHKN,
and 3OHKG—are intrinsically unstable, and under physiological
conditions they undergo spontaneous deamination,11,26 forming
highly reactive carboxyketoalkenes (CKA (IUPAC name: 4-(2-
aminophenyl)-4-oxocrotonic acid), 3OHCKA, and 3OHCKAG,
respectively in Scheme 1). The CKA molecules can covalently
bind to the nucleophilic amino acid residues of proteins (primarily
cysteine, histidine, and lysine).13,27–28 These reactions lead to
disruption of protein functionality and increased susceptibility of
the protein to UV light absorption. It has been demonstrated that
protein-bound kynurenines are better photosensitizers than free
UV filters,29–31 and UV irradiation of modified proteins can lead
to further damage either due to direct photoreaction, or through
aInternational Tomography Center SB RAS, Institutskaya 3a, Novosibirsk,
630090, Russia. E-mail: yura@tomo.nsc.ru
bNovosibirsk Institute of Organic Chemistry SB RAS, Acad. Lavrentjev 9,
Novosibirsk, 630090, Russia
cCaudill Laboratories, Department of Chemistry, CB #3290, University of
North Carolina, Chapel Hill, NC, 27599. E-mail: mdef@unc.edu
† Electronic supplementary information (ESI) available: The vinyl and
aromatic region of the 300 MHz 1H NMR spectrum of CKA with
assignments of the vinyl protons to establish stereochemistry. See DOI:
10.1039/b903196k
2958 | Org. Biomol. Chem., 2009, 7, 2958–2966
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Scheme 1
the formation of reactive oxygen species such as singlet oxygen
and superoxide. The protection of proteins from modification
by deaminated UV filters is provided by reduced glutathione
(GSH),12,13,32 which can bind CKAs, forming new UV filters such
as the 3OHKG-GSH adduct.
demonstrates the importance of steric factors in the addition
reactions. Furthermore, it implies that the addition of CKAs to
the amino acid residues in proteins may be much slower than
their addition to free amino acids. While a consistent picture of
the molecular level reactions related to cataract development is
emerging, there are still several missing pieces of information.
For example, the rates of formation of CKA adducts are known
for many compounds,13 but their rates of decomposition and
eventual disposition have not been determined. The mechanism
of the deamination reactions of KN and 3OHKN has yet to be
elucidated, nor have there been any quantitative measurements of
the kinetics of this important reaction as a function of pH.
In this paper we report the results of a study on the stability of
the “primary” UV filters, KN and 3OHKN, and on the stability
of deaminated KN adducts to several amino acids and GSH. The
results presented below establish the pH dependence for the
reactions of KN and 3OHKN deamination. A mechanism for
the deamination reaction is proposed that is consistent with these
results. Additionally, we have determined the rate constants for the
decomposition of CKA adducts to glutathione, cysteine, histidine,
and lysine under physiological conditions. These reactions will be
discussed and compared with measurements of similar reactions
involving the lens proteins themselves.
Before presenting the results it is important to note the following
with regard to nomenclature in these systems: the adducts to amino
acids and GSH are labeled throughout this paper as KN-Cys,
KN-GSH, etc. However, it is actually the deamination product
(CKA or 3OHCKA) that is covalently binding to the amino acids
or antioxidant. We are following the convention of acronyms
for these adducts first put forward by the research group of
R. J. W. Truscott (Sydney), to avoid confusion, and to facilitate
comparisons of our results with theirs. The structures of all of these
adducts are known and have been previously published.28,33,34
Recently, it has been revealed that kynurenine adducts to indi-
vidual amino acids are also unstable: they decompose, restoring
the CKA.28,33,34 Thus, the modification of the lens proteins by
deaminated UV filters is a complex, dynamic process in which the
same CKA molecule can reversibly add several times to different
amino acid residues and/or antioxidants present in the lens. The
dynamics of UV-filter evolution are governed by many factors.
The UV filters and antioxidants are transported into the lens
nucleus from the cortex and surrounding humor, and the efficiency
of the protective systems strongly depends on the transport of
antioxidants and oxidation products to and from the lens. It
has been reported that an internal barrier is formed between
the lens cortex and nucleus in middle age.21,35,36 The existence
of such a barrier slows down the movement of antioxidants to
the lens center, exposing the lens nucleus to additional oxidative
stress.
The chemical equilibrium between free and covalently bound
UV filters is determined by the rates of non-enzymatic reactions:
kynurenine deamination, addition of CKAs to amino acid residues
and antioxidants, and finally, CKA-adduct decomposition. The
rate constants for CKA addition to nucleophilic amino acids and
antioxidants present in the human lens have been measured.13 The
rates for CKA addition to the thiols cysteine and glutathione,
are several orders of magnitude higher than the rates of addition
to other molecules, including amino acids. In the same study
it was noted that at the physiologically relevant temperature of
◦
37 C, the rate constant of the reaction of CKA with cysteine is
almost 20 times higher than that with glutathione. This result
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 2958–2966 | 2959
©

View Article Online
Results
Decomposition of KN and 3OHKN as a Function of pH
The kinetics of KN decay observed under anaerobic incubation
(37 ◦C) of 1 mM kynurenine in aqueous buffered solutions at
different pH values are shown in Fig. 1. In a control experiment at
pH 7.0, the incubations were performed in non-buffered solution.
The pH value was adjusted by dropwise addition of a NaOH
solution. These solutions give essentially the same result as that
measured in buffered solution. We can conclude that the buffer
components do not contribute to the decomposition reactions,
therefore all further measurements were performed in buffered
solutions. The kinetic data were treated as monoexponential
functions and the solid lines in Fig. 1 correspond to calculated
curves. The pH dependence of the deamination rate constant is
plotted in Fig. 2.
Fig. 2 pH dependence of the first-order rate constant of KN decomposi-
tion during the incubation at T = 37 ◦C.
and 3OHKN yellow. Kinetic curves for 3OHKN decomposition
are shown in Fig. 3, and the pH dependence of the 3OHKN
deamination rate constant is plotted in Fig. 4. The pH dependences
for KN and 3OHKN have similar shapes: the deamination rate
is almost pH independent below pH 8, above which it steeply
increases. Note that the absolute rate increase for 3OHKN at high
pH is much higher than for KN.
Fig. 1 Kinetics of KN thermal decomposition under anaerobic condi-
tions for different pH values: ꢀ - pH 6.6; ᭜ - pH 8.5; ᭺ - pH 9; ꢀ- pH 9.7
during the incubation at T = 37 ◦C. Solid lines show monoexponential
fits.
It was reported earlier11,26 that the major pathway of KN
thermal decomposition is deamination, which we have confirmed
by analysis of the HPLC profile of our incubated samples. The
major products found after KN incubation are carboxyketoalkene
CKA (the direct deamination product), KN yellow from CKA
cyclization, and the KN dimerization product (Scheme 2). The
identity of these products was confirmed by HPLC with UV–VIS
and mass spectrometric detection.
Fig. 3 Kinetics of 3OHKN thermal decomposition under anaerobic
conditions at different pH values: ᭹ - pH 5.5; ᭺ - pH 6.8; ꢀ - pH 8.5
during the incubation at T = 37 ◦C. Solid lines show exponential fits.
Decomposition of K-Lys, KN-His, KN–Cys, and
KN-GSH Adducts
We made similar measurements to study the degradation of
3OHKN. The HPLC analysis revealed the formation of 3OHCKA
The kinetics of the decomposition of deaminated KN adducts to
lysine, histidine, cysteine, and glutathione during incubation under
Scheme 2
2960 | Org. Biomol. Chem., 2009, 7, 2958–2966
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Table 1 Rate constants of adduct decompositions (kdec) and addition of
amino acids to deaminated kynurenine addition to amino acids (kadd) under
physiological conditions (T = 37 ◦C, pH 7.0)
Adduct
k
dec/s-1
k
add/M-1s-1a
KN-Lys
KN-His
KN-NAc-His
KN-Cys
KN-GSH
<10-7
(2.3 0.6) ¥ 10-4
(1.2 0.3) ¥ 10-4
—
(3.4 1.0) ¥ 10-7
(3.8 1.2) ¥ 10-7
(1.7 0.4) ¥ 10-5
(5.8 0.7) ¥ 10-6
36
4
2.1 0.2
a from ref. 13.
6.6 ¥ 10-5 M, Fig. 5A), which is within the experimental error.
For the KN-His adduct, the level of decomposition was somewhat
higher, and after 460 h of incubation, almost half of the starting
material had decomposed (Fig. 5B). It has been reported34 that the
protection of the side-chain amino group of adducts by the tert-
butoxycarbonyl (t-Boc) group significantly raises adduct stability
twofold to threefold.
To verify this, we incubated an additional adduct in which
the a-amino group of the His side chain was protected with
an N-acetyl group, KN-NAc-His. The kinetics of KN-NAc-His
decomposition was essentially the same as that for the unprotected
Fig. 4 pH dependence of the first-order rate constant of 3OHKN
decomposition during the incubation at T = 37 ◦C.
physiological conditions (T = 37 ◦C, pH 7.0) are shown in Fig. 5.
The solid lines in Figs. 5B, 5C, and 5D are good monoexponential
fits, and the calculated rate constants of the adduct decompositions
are listed in Table 1.
After 200 h of incubation, the KN-Lys adduct showed only
about 7% loss of the starting material (from 7 ¥ 10-5 to
Fig. 5 Kinetics of adduct decomposition during the incubation under physiological conditions (T = 37 ◦C, pH 7.0). A) Decay of KN-Lys adduct.
B) Decay of KN-His (ꢀ) and KN-NAc-His (᭺) adducts. C) Decay of KN-Cys (ꢀ); formation of CKA (᭺), and formation of KN yellow (D). D) Decay
of KN-GSH (ꢀ); formation of CKA (᭺), and formation of KN yellow (ꢀ). In all plots, solid lines show exponential fits. See text at the end of the
Introduction for definition of acronyms and structures.
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 2958–2966 | 2961
©

View Article Online
KN-His adduct (Fig. 5B). It is likely that the bulky t-Boc group
sterically hinders the hydrolysis of the KN side chain, and thus
decelerates the overall rate of adduct decomposition.34 The smaller
N-acetyl group does not significantly influence the rate of adduct
decay.
et al., where measurements were made at two pH values.11 Under
physiological conditions (37 ◦C, pH 7) they found that after 7 days
of incubation, the losses of starting material of KN, 3OHKN,
and 3OHKG were 58, 79, and 70%, respectively, which roughly
corresponds to deamination rate constants of 1.4 ¥ 10-6 s-1, 2.7 ¥
10-6 s-1, and 2 ¥ 10-6 s-1. In basic solution (pH 9) the rate of
decomposition for all three UV-filter compounds increased.11 The
deamination rate constants for neutral solutions, obtained in the
present quantitative study (1.2 ¥ 10-6 s-1 for KN and 4.3 ¥ 10-6 s-1
for 3OHKN), are in a good agreement with these previous results.
For both kynurenines studied here, the pH dependences of the
deamination rate constants show two regions: below pH 8 the
deamination rate is weakly pH-dependent, while in basic solutions
the rate constants of deamination continuously grow with pH
(Figs. 2 and 4). This indicates that the deamination reactions of UV
filters at low and high pH values proceed by different mechanisms.
It should also be noted that, for all pH values, the deamination of
3OHKN is faster than that of KN by a factor of three to five.
The fastest decomposition was observed for cysteine and
glutathione adducts of KN (KN-SR) (Figs. 5C and 5D): the
calculated decomposition rate constants for these compounds are
higher than those for the KN-His adducts by more than an order
of magnitude (Table 1). It is interesting that the rate constant
for KN-Cys decay is almost threefold higher than that for KN-
GSH. Perhaps this effect can be explained in the same way as the
influence of the t-Boc group: the bulkier GSH moiety obstructs
the access of hydroxyl ions to the reaction center in the adduct.
The HPLC profiles of the incubated samples demonstrate that
the major products of adduct decomposition are carboxyke-
toalkene (CKA), KN yellow, and KN dimer; some minor unidenti-
fied KN-derived products have also been observed. The kinetics of
CKA and KN yellow formation, observed during the incubation
of KN-Cys and KN-GSH adducts, are shown in Figs. 5C and 5D,
respectively. The formation of KN yellow is much slower than that
of CKA, which suggests that the formation of these compounds
is stepwise: KN yellow is the product of CKA cyclization,11,37,38
occurring at 37 ◦C with the rate constant kcycl = (1.6 0.4) ¥
10-6 s-1 (Scheme 3).13
Mechanism of deamination
The deamination reactions of KN and 3OHKN take place by
simultaneous loss of ammonia from the a-carbon and a proton
from the b-carbon (Scheme 4). The overall driving force for these
reactions may arise from the fact that the final products are a,b-
unsaturated keto-acids containing an extended p-bond system not
present in the starting materials. There are three major factors that
can influence the rate of the deamination reaction: 1) the acidity
of the b-proton; 2) the strength of the base removing this proton,
and 3) the leaving group’s base strength (weaker bases make better
leaving groups). At most of the pH values studied in this work, the
a-amino group is protonated. This makes it a reasonable choice as
the leaving group (as NH3) in the elimination step. Deamination
is not possible by an elimination pathway from the deprotonated
structure, because NH2- is a very strong base and therefore a very
poor leaving group.
Scheme 3
It is logical to suggest that the decomposition of the adducts
results in the formation of CKA and free amino acids. However,
thiols Cys and GSH add to the unsaturated double bond of
CKA at fast rates (Table 1), restoring the starting adducts in both
cases. Thus, the quasi-equilibrium concentration of CKA during
incubation is actually much lower than observed in the experiment.
We have assumed that the decomposition of adducts KN-Cys
and KN-GSH occurs with the formation of the oxidized thiols,
Cys-Cys and GSSG. The presence of free thiols in KN-Cys and
KN-GSH solutions after incubation was examined with the use of
the Ellman’s reagent DTNB.39 No free thiol groups were detected
in the incubated adduct solutions. However, examination of the
incubated KN-GSH solutions using a MALDI-TOF mass spec-
trometer revealed the presence of oxidized glutathione, GSSG. The
mechanism of formation of the oxidized thiols is unclear. In a
recent paper by Parker et al.34 it was reported that the addition of
GSH to KN-t-Boc-Cys solutions promotes the decomposition of
the adduct. It is possible that reduced thiols (RSH), formed due
to breakdown of the starting adducts, can react with the adducts,
yielding CKA and the oxidized thiols, RSSR. However, the effect
of GSH on the rate of adduct breakdown became pronounced
only with GSH concentrations above 10 mM,34 which is much
higher than we would ever observe via adduct breakdown under
our experimental conditions.
It should be noted that NH3 still has significant base strength
and really isn’t an ideal leaving group for an elimination reaction,
-
but it is a much better leaving group than NH2 so it is through
this pathway that the reaction must proceed. The “tepid” leaving-
group ability of NH3 is most likely why the overall reaction rates
for deamination are quite slow for these compounds. Of course,
the advantage of having an NH3 leaving group is that it will
+
immediately be protonated to NH4 (pKa = 9.3) in most of the
solutions studied here. The positively charged ion is completely
non-nucleophilic, making the back reaction impossible. This is
an important factor in the kinetic stability of the deamination
products. An additional important point is that the leaving group
is the same for deamination of KN and 3OHKN, therefore the
observed differences in reactivity between these compounds must
be based on other criteria. We will discuss this issue in more detail
below.
Base strength is a major factor in these reactions because at
high pH there is a strong base present (HO-), while at low pH
only water is present, a weak base. This by itself may explain
the acceleration in the deamination rate seen for both KN and
3OHKN above pH 8, i.e. we are observing generalized base
catalysis. A question then arises as to the more precise nature
of the elimination mechanism. Obviously we can rule out the
Discussion
pH Dependence of KN and 3OHKN degradation
The degradation of UV-filter compounds in neutral and basic
solutions has been studied in a qualitative fashion by Taylor
2962 | Org. Biomol. Chem., 2009, 7, 2958–2966
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
Scheme 4
E1 extreme limit as we are in basic solution where carbocations
will not form. An E2 mechanism is unlikely at all pH values
because we expect the leaving group ability of the amino group
to be very different in its protonated vs. deprotonated states. For
these reasons, we have drawn the mechanism in Scheme 4B as
“E1CB”, that is, deprotonation at the b-position takes place before
the leaving group leaves, creating the enolate (conjugate base,
CB). Of course it is possible to have an “E1CB-like” transition
state in an E2 reaction, which cannot be ruled out. Either way,
this deprotonation step is the basis for the observed general
base catalysis (acceleration of KN decomposition from pH 8 to
pH 10.5).
The observed rate constant for deamination is not directly
proportional to the HO- concentration in solution: the increase
in the rate of decay is only a factor of 20 even though the HO-
concentration has increased by a factor of 300 (2.5 pH units).
Furthermore, the rate constants for KN decomposition at pH 10.5
and pH 11.2 are almost identical, meaning that the acceleration of
KN decomposition tails off (Fig. 2, last data point). We attribute
this to the deprotonation of the amino group, rendering it a much
poorer leaving group. Both of these observations point to an E2
mechanism where the transition state is more like an E1CB reaction,
i.e., deprotonation at the b-carbon is almost complete before
the leaving group on the a-carbon begins to leave. An aqueous
solution of pH 11.2 is a rather harsh environment for any organic
molecule, therefore it should be noted that other degradation
processes may also be taking place under these conditions. Still,
the data in Fig. 2 strongly support every aspect of the mechanism
proposed in Scheme 4B.
(slow), leading to loss of NH3 and reformation of the protonated
carbonyl. Loss of a proton from the carbonyl oxygen is the last
(fast) step, leading to the deamination product. Although the
initial protonation is not highly favored, because of the low H3O+
concentration, it should be remembered that our overall reaction
rates are very slow compared to “normal” organic flask reactions.
On the time scale of our observations (sometimes hundreds of
hours) it seems likely that even moderately acidic solutions will
eventually lead to protonation of the carbonyl groups in KN and
3OHKN, with subsequent generalized acid catalysis of the E2
reaction.
At high pH values, deprotonation to make the enolate anion
directly from the g-carbonyl can be expected when there is a
significant concentration of HO- (Scheme 4B). The elimination
reaction can proceed immediately from the enolate anion, al-
though in aqueous solution the enol may also be present in a rapid
equilibrium with the anion. Since the final product is the same for
either structure, the elimination step is shown at the bottom of
Scheme 1B only for the enolate anion. The elimination reaction
produces exclusively the E-isomer of the deamination product, as
determined from NMR data (see the ESI†), where the vinyl region
shows two doublets at 6.65 ppm and 7.57 ppm, with a typical trans
2J coupling constant of 15.8 Hz.
We propose that at all pH values, either the enol or enolate
anions of KN and 3OHKN are involved in their deamination
reactions. Formation of the enols is catalyzed by excess acid
(hydronium ion) and formation of the enolate anions is catalyzed
by excess base (hydroxide ion). Scheme 4 summarizes the mecha-
nism at A) low pH values and B) high pH values. The proposed
mechanism accounts for two very important features of the kinetic
data: first and foremost, it gives clear reasoning for the acceleration
of the deamination reaction at high pH values, i.e. when there is a
significant concentration of strong base and there is an enolizable
proton present. Second, the mechanism accounts for the fact that
at pH values below 8, deamination is still possible but is slower and
must involve the enol of the g-carbonyl because water is not basic
enough to participate directly in the elimination reaction. Since the
leaving group is non-ideal at either pH value, base strength is the
Concerning the acidity of the b-protons of KN and 3OHKN, the
additional carbonyl group in the g-position may greatly influence
the rate of deprotonation at the b-site. At low pH values, with water
as the base, no direct b-deprotonation is expected. However, excess
hydronium ions can protonate the carbonyl oxygen (Scheme 4A,
first step). The protonated carbonyl, through an inductive effect,
makes the b-protons much more acidic and water can participate
in the second step of the mechanism to make the enol. The enol
can then undergo the electron-pair shift illustrated in the third step
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 2958–2966 | 2963
©

View Article Online
determining factor in the observed trend in the rates of reaction
as a function of pH.
of post-translational modifications of the lens proteins by UV
filters.23,24,28,33,34,41,42 It has been demonstrated that the initial
attachment sites of deaminated KN and deaminated 3OHKN to
proteins are Cys residues.24 However, other studies showed that
crystallins isolated from human lenses contain significant levels
of bound kynurenine, and the modifications occur mostly via His
and Lys residues, with only minor levels of kynurenine attached
to Cys residues.33 Thus, the covalent attachment of deaminated
kynurenines to antioxidants such as GSH, amino acids, and
proteins appears to be a reversible reaction, and the stabilities
of the adducts play a more important role in this chemistry than
the rate constants of the addition reactions.
The concentrations of KN-Lys and KN-His adducts in human
lens proteins significantly increase with age.28 This impact of
crystallin modification by UV filters is manifested in lens col-
oration, increases in fluorescence intensity, and opacification. It
is interesting that the levels of protein-bound kynurenines in the
nuclei of cataractous lenses are much lower than in normal ones.42
This probably means that protein-bound UV filters are in dynamic
equilibrium with free ones, and the decrease in concentration of
protein-bound kynurenines reflects the total decrease of UV filter
levels in cataractous lenses.42
The difference in reactivity between KN and 3OHKN is worthy
of discussion. Because the OH group in 3OHKN is meta to
the g-carbonyl substituent, the lone pairs on the OH oxygen
are unable to donate into the phenyl ring. For this reason
the OH lone pairs cannot participate in any stabilization or
destabilization via resonance effects in the reactants, products or
reactiveintermediates. However, there maybe an inductive effectof
the phenolic oxygen atom through the s-bond framework, pulling
electron density away from the g-carbonyl carbon. This in turn
will make the neighboring b-protons slightly more acidic, meaning
that, for the same reaction conditions, 3OHKN should degrade
faster than KN. This is exactly what is observed experimentally.
It should be noted, however, that the pKa of 2-aminobenzoic
acid (4.96) is lower than that of 2-amino-3-hydroxybenzoic acid
(5.19), which would seem to contradict our argument for an
inductive effect. However, the benzoic acids may not be good
models for the elimination (or enolization) reaction because there
may be significant intramolecular H-bonding to the amino group
through a six-membered-ring transition state. This H-bonding will
be competitive between the OH on the benzoic acid and OH on
the phenyl ring, as shown in Scheme 5. Such competition will
reduce the ability of the amino group to assist in the deprotonation
reaction, which would explain the rather significant difference in
pKa values of the two benzoic acids.
Comparison with KN-protein incubation studies
In a previous study we measured the rate constants of deami-
nation product (CKA) linkage to nucleophilic amino acids and
antioxidants, and showed that at physiological conditions the rate
constant of CKA binding to Cys is 5 orders of magnitude higher
than its value for His or Lys.13 An earlier study of the incubation
of crystallins in the presence of KN and KN derivatives has been
reported,34 and the Cys residue in these proteins was determined
to be initial site of chemical modification. Despite lower rate
constants for CKA binding to Lys and His, modifications of these
residues were also revealed during incubated protein analysis.
According to Vazquez et al.,28 the KN-Cys adduct decomposes
under physiological conditions, yielding deaminated KN and KN
yellow, and our experimental data match these findings.
Experiments on KN incubation with crystallins have been
reported27,28 and tryptic digestion analysis of these proteins
showed Cys, His and Lys residues as the preferred sites of KN
modification. Longer incubation of KN with crystallins results
in an increase in the presence of the more stable KN-His and
KN-Lys adducts; this correlates with our findings on adduct
stability. Vazquez et al.28 have reported that KN-His and KN-Lys
concentrations appear to be a function of time and increase during
aging. The concentration of Cys-modified residues decreases with
age according to adduct instability, however modification on His
residues accumulates.
Scheme 5
Competition for H-bonding is expected to be a minor effect
in the right hand structure (Scheme 5) leading to the enolate in
kynurenine degradation chemistry. Because of the large difference
in pKas of the acids and ketones, one can use the Hammond
postulate to predict a much later transition state for deprotonation
of the ketone. The amino group will therefore play a smaller
role in the deprotonation reaction, rendering intramolecular
H-bonding effects much less important. Correspondingly, the
electron withdrawing inductive effect of the phenolic OH oxygen
may be more pronounced in our system. This interplay of two
intramolecular effects is an interesting problem that should be of
interest from a computational perspective in future work.
Kinetic analysis
It is interesting to note that in tryptic digests of aged lens crys-
tallins, modified Cys residues were detected in lesser concentration
than those for Lys or His, despite higher rate constant values for
KN-Cys adduct formation.28 This was explained by proposing that
the release of CKA molecules may allow them to react with other
amino acid residues, causing additional modification in proteins. It
is expected that CKA bound to Cys residues will be released during
prolonged thermolysis, and modifications on His and Lys will
accumulate. However, even after the termination of incubation, a
significant amount of CKA remained bound to Cys residues. From
Table 1 summarizes the data for KN adduct stability, as well as the
rate constants for adduct formation, which occurs via the Michael
addition reaction40 of CKA to amino acids. Analysis of the data
shows that KN adducts to thiols—Cys and GSH—are significantly
less stable than those to His and Lys; at the same time, the rate
constants of Cys and GSH addition to CKA are several orders of
magnitude higher than those of other amino acids/antioxidants.
These results help to explain the results of protein incubation in
the presence of KN. They also help to understand the analysis
2964 | Org. Biomol. Chem., 2009, 7, 2958–2966
This journal is
The Royal Society of Chemistry 2009
©

View Article Online
these results it appears that CKA attached to Cys residues may be
hidden in the protein tertiary structure and become inaccessible
for hydrolysis. This result certainly confirms that there is a strong
protein environmental effect on the properties of KN derived
compounds. It is possible to assume that during aging, the process
of KN-Cys adduct cleavage may shift the CKA adduct distribution
in favor of the formation of KN-Lys and KN-His adducts. This
possibility will be explored in future experiments.
It has been proposed that another nucleophile, GSH, is the
major antioxidant in the lens-inhibiting CKA binding to the lens
proteins.5,12,32,43 The rate constant for the addition of CKA to GSH
is rather high compared to the formation of KN-His or KN-Lys.13
However, in the work presented here, we have found that the KN-
GSH adduct turned out to be less stable than other CKA adducts.
Unbound GSH may act as a target for CKA attack, but taking
into consideration the rate constant for KN-GSH adduct cleavage
(Table 1) and the decrease of GSH concentration in the lens
during aging, reactions involving GSH diminish lens protection
overall. In fact, insufficient levels of GSH in the lens will lead
to an accumulation of Lys- and especially His-modifications in
crystallins.28 It was recently established5 that GSH attached to
3OHKG is the main UV-filter-derived compound present in the
lens, which is definitive proof of the protective function of this
antioxidant.
salt solutions (5.4 mM) containing a tenfold molar excess of Lys,
His, NAc-His, Cys, or GSH were prepared using phosphate buffer
of pH 9.5. The pH value was adjusted by drop-wise addition of
NaOH if required and controlled by an Orion Research pH meter
with a glass electrode. The pH value remained constant during the
entire incubation period. The solution (volume 20 mL) was placed
in a glass vial, capped, bubbled with argon, sealed, and incubated
in a thermostatted water bath at 37 ◦C for 48 h. After incubation,
the resulting mixture was filtered, the pH value adjusted to a
value between 4 and 5 by addition of 1 M hydrochloric acid.
Sample separation was performed by HPLC. Collected fractions
were lyophilized and weighed. After lyophilization the KN-Lys
(yield 60%), KN-His (yield 40%), KN-NAc-His (yield 40%),
KN-Cys (yield 50%), and KN-GSH (yield 80%) adducts were
stored at 4 ◦C.
Incubation
KN, 3OHKN, KN-Lys, KN-His, KN-NAc-His, KN-Cys, and
KN-GSH solutions were prepared in phosphate buffer (0.1 M).
The pH values were controlled by an Orion Research pH meter
with a glass electrode and adjusted by drop-wise addition of
NaOH or hydrochloric acid if required. The solutions were placed
into glass vials (1 mL), capped, bubbled with argon, sealed and
incubated at 37 ◦C. For each pH value a fresh solution was
prepared, and for each incubation time an individual vial was
used. All vials were placed in the thermostat simultaneously and
removed at various intervals. After incubation, the samples were
stored at 4 ◦C and then analyzed by HPLC.
Conclusions
The pH dependences for the rate constants of KN and 3OHKN
deamination show two distinct ranges—strongly pH-dependent
above pH 8 and weakly pH-dependent below pH 7. The deamina-
tion reaction of these compounds accelerates in basic solution,
and the rate constants for 3OHKN are higher than that for
KN. The deamination mechanism is proposed to proceed through
the formation of the enol (low pH) or enolate anion (high pH)
at the aryl carbonyl group of KN and 3OHKN. Amino acid and
antioxidant adducts formed after deamination of KN cleave with
drastically different rates: KN-Cys and KN-GSH are less stable
adducts compared to KN-His or KN-Lys. Under physiological
conditions, modifications at Lys and His residues accumulate in
the lens, however the mechanisms when crystallin proteins are
involved are not clearly understood and remain under active
investigation.
Separation and analysis
HPLC separation of the amino acid and antioxidant adducts to
deaminated KN was performed with the use of an Agilent LC
1200 chromatograph equipped with an automatic gradient pump
and a multiple wavelength UV–Vis detector. A 9.4 ¥ 250 mm
ZORBAX Eclipse XBD-C18 Semi-Preparative column was eluted
with an acetonitrile/0.05% (v/v) TFA in H2O gradient. The ace-
tonitrile percentage in the gradient was 0–30% (0–2 min), 30–55%
(2–32 min), 55–100% (32–34 min), 100% (34–40 min). The flow rate
was 0.9 mL min-1, and the detection was performed simultaneously
at five wavelengths—254, 290, 315, 360 and 410 nm.
HPLC analyses of the products of thermal degradation of
KN and 3OHKN were performed using an Agilent LC 1100
chromatograph equipped with an quaternary pump, an autosam-
pler, and a diode array detector. Separations were performed
on a 4.6 ¥ 150 mm ZORBAX Eclipse XBD-C8 column using
a methanol/0.1% (v/v) TFA in a H2O gradient. The methanol
percentage in the gradient was 10–40% (0–15 min), 40% (15–
20 min), 40–90% (20–25 min). The flow rate was 0.9 mL min-1, and
detection was performed simultaneously at five wavelengths—254,
290, 314, 360 and 406 nm. Chromatograms were recorded and the
peak areas were integrated with the use of the Agilent ChemStation
for Windows.
Experimental
Materials
D,L-Kynurenine (KN), D,L-kynurenine sulfate salt, 3-hydroxy-
kynurenine (3OHKN), D,L-histidine (His), N-acetyl-L-histidine
(NAc-His), D,L-lysine (Lys), D,L-cysteine (Cys), reduced L-gluta-
thione (GSH), 5,5¢–dithio-bis-(2-nitrobenzoic acid) (DTNB), tri-
fluoroacetic acid (TFA) and a-cyano-4-hydroxycinnamic acid
(CHCA) were purchased from Sigma-Aldrich and used as re-
ceived. Water was doubly distilled. Organic solvents (HPLC grade,
Cryochrom, Russia) were used as received.
Free thiol determination (Ellman’s reaction)
Adduct synthesis
Ellman’s reagent (5,5¢–dithio-bis-(2-nitrobenzoic acid), DTNB)
forms an adduct (2-nitro-5-mercaptobenzoic acid, TNB) in the
presence of free thiols. The TNB ionizes to the TNB- anion in
KN adducts to Lys, His, NAc-His, Cys, and GSH were synthesized
according to the procedure described in ref. 28 The KN sulfate
This journal is
The Royal Society of Chemistry 2009
Org. Biomol. Chem., 2009, 7, 2958–2966 | 2965
©

View Article Online
neutral and basic aqueous solutions, with a relatively intense
absorbance at 412 nm (extinction coefficient 1.4 ¥ 104 M-1cm-1.44,45
The reaction is rapid and stochiometric, and the threshold of
sensitivity is 0.6 nmol mL-1.46 Determination of the quantity of
free GSH or Cys released during the incubation of KN-GSH
11 L. M. Taylor, J. A. Aquilina, J. F. Jamie and R. J. W. Truscott, Exp. Eye
Res., 2002, 75, 165–175.
12 L. M. Taylor, J. A. Aquilina, J. F. Jamie and R. J. W. Truscott, Exp. Eye
Res., 2002, 74, 503–511.
13 L. V. Kopylova, O. A. Snytnikova, E. I. Chernyak, S. V. Morozov and
Yu. P. Tsentalovich, Exp. Eye Res., 2007, 85, 242–249.
14 O. A. Snytnikova, A. Zh. Fursova, E. I. Chernyak, V. G. Vasiliev, S. V.
Morozov, N. G. Kolosova and Yu. P. Tsentalovich, Exp. Eye Res., 2008,
86, 951–956.
15 H. Bloemendal, W. W. de Jong, R. Jaenicke, N. H. Lubsen, C. Slingsby
and A. Tardieu, Prog. Biophys. Mol. Biol., 2004, 86, 407–485.
16 K. J. Dilley and A. Pirie, Exp. Eye Res., 1974, 19, 59.
17 S. Lerman and R. Borkman, Ophthalmic Res., 1976, 8, 335–353.
18 R. J. W. Truscott and R. C. Augusteyn, Biochim. Biophys. Acta, 1977,
492, 43–52.
and KN-Cys adducts was performed by addition of 2 ¥ 10-3
M
of DTNB. The solutions (pH 7) were kept for 20 min at room
temperature, after which the optical absorption at 412 nm was
measured.
Mass spectrometry, NMR, and UV–Vis analysis
Mass spectra of the adducts were obtained on an Agilent
1100 Series LC/MS instrument with a photodiode array and
mass detectors. Data analysis was carried out using the Agilent
ChemStation program. For the detection of oxidized glutathione
(GSSG) in incubated solutions of KN-GSH adducts, mass spectra
were acquired on an Autoflex II Bruker Daltonics MALDI-TOF
Spectrometer in positive-ion reflector mode. The samples were
prepared with the use of a saturated solution of CHCA matrix
(70% acetonitrile, 0.1% TFA). 1H NMR spectra were recorded on
an Avance-300 Bruker NMR Spectrometer. UV–Vis spectra were
obtained on an Agilent 8453 UV-visible Spectroscopy System.
19 N. T. Yu, C. B. Barron and J. F. R. Kuck, Exp. Eye Res., 1989, 49, 189.
20 J. J. Harding, Cataract, Biochemistry, Epidemiology and Pharmacology,
Chapman and Hall , London, 1991.
21 R. J. W. Truscott, Exp. Eye Res., 2005, 80, 709–725.
22 B. D. Hood, B. Garner and R. J. W. Truscott, J. Biol. Chem., 1999, 274,
32547–32550.
23 B. Garner, D. C. Shaw, R. A. Lindner, J. A. Carver and R. J. W. Truscott,
Biochim. Biophys. Acta, 2000, 1476, 265–278.
24 J. A. Aquilina and R. J. W. Truscott, Biochim. Biophys. Acta, 2002,
1596, 6–15.
25 R. J. W. Truscott, Int. J. Biochem. Cell Biol., 2003, 35, 1500–1504.
26 Yu. P. Tsentalovich, O. A. Snytnikova, M. D. E. Forbes, E. I. Chernyak
and S. V. Morozov, Exp. Eye Res., 2006, 83, 1439–1445.
27 J. A. Aquilina and R. J. W. Truscott, Biochem. Biophys. Res. Commun.,
2000, 276, 216–223.
28 S. Vazquez, J. A. Aquilina, J. F. Jamie, M. M. Sheil and R. J. W. Truscott,
J. Biol. Chem., 2002, 277, 4867–4873.
Acknowledgements
29 N. R. Parker, J. F. Jamie, M. J. Davies and R. J. W. Truscott, Free Radical
Biol. Med., 2004, 37, 1479.
30 P. S. Sherin, Yu. P. Tsentalovich, O. A. Snytnikova and R. Z. Sagdeev,
J. Photochem. Photobiol., B, 2008, 93, 127.
31 J. Mizdrak, P. G. Hains, R. J. W. Truscott, J. F. Jamie and M. J. Davies,
Free Radical Biol. Med., 2008, 44, 1108.
32 B. Garner, S. Vazquez, R. Griffith, R. A. Lindner, J. A. Carver and
R. J. W. Truscott, J. Biol. Chem., 1999, 274, 20847–20854.
33 S. Vazquez, N. R. Parker, M. M. Sheil and R. J. W. Truscott, Invest.
Ophthalmol. Visual Sci., 2004, 45, 879–883.
34 N. R. Parker, A. Korlimbinis, J. F. Jamie, M. J. Davies and R. J. W.
Truscott, Invest. Ophthalmol. Visual Sci., 2007, 48, 3705–3713.
35 M. H. Sweeney and R. J. W. Truscott, Exp. Eye Res., 1998, 67, 587–595.
36 B. A. Moffat, K. A. Landman, R. J. W. Truscott, M. H. J. Sweeney and
J. M. Pope, Exp. Eye Res., 1999, 69, 663–669.
This work was supported by the following agencies: FASI state
contract 02.512.11.2278, Russian Foundation for Basic Research
Projects 08-03-00539 and 07-03-00253, President of the Russian
Foundation Scientific School (grant # 3604.2008.3), and the
Division of Chemistry and Material Science, Russian Academy of
Sciences. OAS thanks the Russian Science Support Foundation.
MDEF acknowledges the support of the US National Science
Foundation (grant # CHE-0809530) and the J. W. Fulbright
Scholar Award Program. We thank Professor J. S. Johnson and one
of the referees for helpful comments regarding the deamination
mechanism.
37 T. Tokuyama, S. Senoh, Y. Hirose and T. Sakan, J. Chem. Soc. Jpn.,
1958, 79, 752–761.
38 T. Tokuyama, S. Senoh, T. Sakan, K. S. Brown and B. Witkop, J. Am.
Chem. Soc., 1967, 89, 1017–1021.
39 G. L. Ellman, Arch. Biochem. Biophys., 1959, 82, 70–77.
40 M. B. Smith, and J. March, March’s Advanced Organic Chemistry:
Reactions, Mechanisms, and Structure, Wiley-Blackwell , New York,
2007, p. 612 ff.
References
1 R. van Heyningen, Ciba Found. Symp., 1973, 19, 151–171.
2 A. M. Wood and R. J. W. Truscott, Exp. Eye Res., 1993, 56, 317–325.
3 A. M. Wood and R. J. W. Truscott, Vision Res., 1994, 34, 1369–1374.
4 R. J. W Truscott, A. M. Wood, J. A. Carver, M. M. Sheil, G. M.
Stutchbury, J. Zhu and G. W. Kilby, FEBS Lett., 1994, 348, 173.
5 L. M. Bova, M. H. Sweeney, J. F. Jamie and R. J. W. Truscott, Invest.
Ophthalmol. Visual Sci., 2001, 42, 200–205.
41 A. Korlimbinis and R. J. W Truscott, Biochemistry, 2006, 45, 1950–
1960.
6 I. M. Streete, J. F. Jamie and R. J. W. Truscott, Invest. Ophthalmol.
Visual Sci., 2004, 45, 4091–4097.
7 J. Mizdrak, P. G. Hains, D. Kalinowski, R. J. W. Truscott, M. J. Davies
and J. F. Jamie, Tetrahedron, 2007, 63, 4990–4999.
8 F. Moroni, Eur. J. Pharmacol., 1999, 375, 87.
9 L. M. Bova, A. M. Wood, J. F. Jamie and R. J. W. Truscott, Invest.
Ophthalmol. Visual Sci., 1999, 40, 3237–3244.
42 A. Korlimbinis, J. A. Aquilina and R. J. W. Truscott, Exp. Eye Res.,
2007, 85, 219–225.
43 V. N. Reddy and F. J. Giblin, Ciba Found. Symp., 1984, 106, 65–87.
44 P. W. Riddles, R. L. Blakeley and B. Zerner, Anal. Biochem., 1979, 94,
75–81.
45 P. W. Riddles, R. L. Blakeley and B. Zerner, Methods Enzymol., 1983,
91, 49–60.
10 L. M. Taylor, J. A. Aquilina, R. H. Willis, J. F. Jamie and R. J. W.
Truscott, FEBS Lett., 2001, 509, 6–10.
46 C. K. Riener, G. Kada and H. J. Gruber, Anal. Bioanal. Chem., 2002,
373, 266–276.
2966 | Org. Biomol. Chem., 2009, 7, 2958–2966
This journal is
The Royal Society of Chemistry 2009
©