Proton transfer and carbon-carbon bond cleavage in the elimination of indole catalyzed by Escherichia coli tryptophan indole-lyase

Article abstract of DOI:10.1021/ja991647q

Tryptophan indole-lyase from Escherichia coli catalyzes the reversible cleavage of L-tryptophan to indole and ammonium pyruvate. This reaction is mechanistically interesting since it involves the elimination of an aromatic carbon leaving group. We have been studying the mechanism of tryptophan indole-lyase using rapid-scanning stopped-flow spectrophotometry. Recently, we demonstrated that the rate constant for α-aminoacrylate intermediate formation from α-²H-L-tryptophan exhibits an isotope effect of 3.0 (Sloan, M. J.; Phillips, R. S. Biochemistry 1996, 35, 16165-16173). We have confirmed this previous result (^Dk = 2.99 ± 0.30) and we have now found that β,β-di-²H-L-tryptophan also exhibits a secondary isotope effect (^Dk = 1.17 ± 0.03) on the elimination reaction. Furthermore, α,β,β-tri-²H-L-tryptophan exhibits a multiple isotope effect (^Dk = 4.42 ± 0.67) on the elimination of indole. In addition, there is a significant solvent isotope effect (^Dk = 1.79 ± 0.11) on indole elimination in D₂O. This solvent isotope effect combines with the effect of α-deuterium, since elimination of α-²H-L-tryptophan in D₂O exhibits ^Dk = 4.30 ± 0.16. In addition, the rate constant for indole elimination shows a linear Eyring plot between 5 and 35 °C. In the direction of tryptophan synthesis, the reaction of the α-aminoacrylate intermediate with indole to form a quinonoid intermediate also exhibits a kinetic isotope effect for 3-²H-indole, with ^Dk = 1.88 ± 0.19. In contrast to our expectations, the results suggest that the proton transfer and carbon-carbon bond cleavage in the elimination reaction are very nearly simultaneous and that the indolenine structure is a transient intermediate which occupies a very shallow well on the reaction coordinate, or a transition state, in the reaction of Trpase.

Full text of DOI:10.1021/ja991647q

1008
J. Am. Chem. Soc. 2000, 122, 1008-1014
Proton Transfer and Carbon-Carbon Bond Cleavage in the
Elimination of Indole Catalyzed by Escherichia coli Tryptophan
Indole-Lyase
Robert S. Phillips,*,†,‡,§ Bakthavatsalam Sundararaju,† and Nicolai G. Faleev|
Contribution from the Department of Chemistry, Department of Biochemistry and Molecular Biology, and
Center for Metalloenzyme Studies, UniVersity of Georgia, Athens, Georgia 30602-2556, and NesmeyanoV
Institute of Elemento-organic Compounds, Russian Academy of Sciences, Moscow, Russia
ReceiVed May 17, 1999
Abstract: Tryptophan indole-lyase from Escherichia coli catalyzes the reversible cleavage of L-tryptophan to
indole and ammonium pyruvate. This reaction is mechanistically interesting since it involves the elimination
of an aromatic carbon leaving group. We have been studying the mechanism of tryptophan indole-lyase using
rapid-scanning stopped-flow spectrophotometry. Recently, we demonstrated that the rate constant for
R-aminoacrylate intermediate formation from R-2H-L-tryptophan exhibits an isotope effect of 3.0 (Sloan, M.
J.; Phillips, R. S. Biochemistry 1996, 35, 16165-16173). We have confirmed this previous result (Dk ) 2.99
( 0.30) and we have now found that â,â-di-2H-L-tryptophan also exhibits a secondary isotope effect (Dk )
1.17 ( 0.03) on the elimination reaction. Furthermore, R,â,â-tri-2H-L-tryptophan exhibits a multiple isotope
effect (Dk ) 4.42 ( 0.67) on the elimination of indole. In addition, there is a significant solvent isotope effect
(Dk ) 1.79 ( 0.11) on indole elimination in D2O. This solvent isotope effect combines with the effect of
D
R-deuterium, since elimination of R-2H-L-tryptophan in D2O exhibits k ) 4.30 ( 0.16. In addition, the rate
constant for indole elimination shows a linear Eyring plot between 5 and 35 °C. In the direction of tryptophan
synthesis, the reaction of the R-aminoacrylate intermediate with indole to form a quinonoid intermediate also
D
exhibits a kinetic isotope effect for 3-2H-indole, with k ) 1.88 ( 0.19. In contrast to our expectations, the
results suggest that the proton transfer and carbon-carbon bond cleavage in the elimination reaction are very
nearly simultaneous and that the indolenine structure is a transient intermediate which occupies a very shallow
well on the reaction coordinate, or a transition state, in the reaction of Trpase.
Introduction
of tryptophan by Hopkins and Cole in 1903.5 In addition to the
elimination reaction shown in eq 1, Trpase can catalyze the
elimination in vitro of a wide range of amino acids with leaving
groups attached to the â-carbon, including L-serine,6 O-alkyl-
L-serines,6 O-acyl-L-serines,7 S-alkyl-L-cysteines,6 S-(o-nitro-
phenyl)-L-cysteine,8 â-chloroalanine,6 and 2,3-diaminopropi-
onate.6 All of these nonphysiological substrates possess chemically
reasonable leaving groups which contain relatively electrone-
gative elements. However, the physiological reaction of Trpase
is mechanistically intriguing, as it requires the elimination of a
formally unactivated aromatic carbon leaving group. Hence,
there must be additional catalysis to activate the indole ring to
make it a suitable leaving group under physiological conditions.
Studies performed with R-2H-L-tryptophan have demonstrated
that there is intramolecular transfer of the R-proton to C-3 of
indole in the reaction.9,10 Previously, we demonstrated that
analogues of the indolenine tautomer of L-tryptophan,11 2,3-
dihydro-L-tryptophan and oxindolyl-L-alanine (Chart 1), are
potent competitive inhibitors of Trpase and tryptophan syn-
thase.12,13 Furthermore, the inhibition of Trpase is stereospecific
Tryptophan indole-lyase (Trpase, EC 4.1.99.1) catalyses the
reversible hydrolytic cleavage of L-tryptophan to form indole
and ammonium pyruvate (eq 1). This enzyme is found in a
number of enteric bacteria including Escherichia coli,1 Proteus
Vulgaris,2 and Haemophilus influenzae.3 The production of
indole during the bacterial putrefaction of protein was first
reported by Bopp in 1849.4 The source of the indole in protein
remained obscure until the synthesis and structure elucidation
* To whom correspondence should be addressed at Department of
Chemistry, University of Georgia. Phone: (706) 542-1996. Fax: (706) 542-
9454. E-mail: rsphillips@chem.uga.edu.
† Department of Chemistry, University of Georgia.
‡ Department of Biochemistry and Molecular Biology, University of
Georgia.
§ Center for Metalloenzyme Studies, University of Georgia.
(5) Hopkins, F. G.; Cole, S. W. J. Physiol. (London) 1903, 29, 451.
(6) Watanabe, T.; Snell, E. E. J. Biochem. (Tokyo) 1977, 82, 733-745.
(7) Phillips, R. S. Arch. Biochem. Biophys. 1987, 256, 302-310.
(8) Suelter, C. H.; Wang, J.; Snell, E. E. FEBS Lett. 1976, 66, 230-
232.
(9) Vederas, J. C.; Schleicher, E.; Tsai, M.-D.; Floss, H. G. J. Biol. Chem.
1978, 253, 5350-5354.
(10) Sloan, M. J.; Phillips, R. S. Biochemistry 1996, 35, 16165-16173.
(11) Davis, L.; Metzler, D. Enzymes 1972, 7, 334-338.
| Russian Academy of Sciences.
(1) Snell, E. E. AdV. Enzymol. Relat. Areas Mol. Biol. 1975, 42, 287-
333.
(2) Kamath, A. V.; Yanofsky, C. J. Biol. Chem. 1992, 267, 19978-
19985.
(3) Martin, K.; Morlin, G.; Smith, A.; Nordyke, A.; Eisenstark, A.;
Golomb, M. J. Bacteriol. 1998, 180, 107-118.
(4) Bopp, F. Justus Liebigs Ann. Chem. 1849, 69, 16-37.
10.1021/ja991647q CCC: $19.00 © 2000 American Chemical Society
Published on Web 01/29/2000

Elimination of Indole
J. Am. Chem. Soc., Vol. 122, No. 6, 2000 1009
Chart 1
for (3R)-2,3-dihydro-L-tryptophan.13 We interpreted these results
as evidence for an indolenine intermediate in the reactions of
both Trpase and tryptophan synthase. In the present work, we
examined the effects of isotopic substitution of tryptophan and
solvent on the pre-steady-state kinetics of the indole elimination
reaction of Trpase to obtain evidence for an indolenine
intermediate. In contrast to our expectations, the results suggest
that the proton transfer and carbon-carbon bond cleavage in
the elimination reaction are very nearly simultaneous and that
the indolenine structure is a transient intermediate which
occupies a very shallow well on the reaction coordinate, or a
transition state, in the reaction of Trpase.
Figure 1. Rapid-scanning stopped-flow spectra of the reaction of E.
coli Trpase with 10 mM L-tryptophan and R-2H-L-tryptophan in the
presence of 5 mM benzimidazole. The scan times are the following:
1, 0.001 s; 2, 0.003 s; 3, 0.005 s; 4, 0.010 s; 5, 0.020 s; 6, 0.040 s; 7,
0.080 s; 8, 0.160 s.
the R-aminoacrylate complex with benzimidazole releases
pyruvate at a considerably slower steady-state rate. The 345
nm intermediate forms concomitant with decay of the external
aldimine (λmax ) 422 nm) and quinonoid (λmax ) 505 nm)
intermediates, with an isosbestic point at 364 nm and clean first-
order kinetics (Figure 1). In contrast, in the absence of
benzimidazole, only an increase in the absorbance of the 422
and 505 nm peaks is observed when Trpase is mixed with
L-tryptophan.10,14 The observed rate constant for formation of
the 345 nm intermediate at 25 °C is 32 ( 2 s-1, and this rate
constant is not significantly affected by the concentration of
either L-tryptophan or benzimidazole, although the amplitude
is increased with [benzimidazole].14 The same rate constant can
be obtained either by fitting the absorbance increase at 345 nm
(Figure 2A), the absorbance decrease at 422 nm (Figure 2B),
or the absorbance decrease in the second phase at 505 nm
(Figure 2C). The absorbance changes at 345 nm show more
noise than those at 422 or 505 nm due to the lower output of
the Xe lamp in the near-UV region and to the reduced efficiency
of the visible gratings in the stopped-flow instrument below
350 nm (see Figure 1). However, the noise level of the data
obtained at 345 nm does not significantly affect the results of
fitting. For this study, rate constants were obtained from single
exponential fitting of the first 200 ms of the reaction at either
345 or 422 nm, as fitting of the 505 nm time courses can be
problematic due to the rapid rise-fall behavior (Figure 2C).
These data at either 345 or 422 nm can be fit to a single
exponential process with very high precision. At longer times,
the 345 nm absorbance peak slowly decreases (Figure 2A),
concomitant with an increase in absorbance at 422 nm (Figure
2B) and 505 nm (Figure 2C), due to the slowly increasing
[indole] which competes with benzimidazole for the R-amino-
acrylate intermediate.
Results
r-Deuterium Isotope Effect on Indole Elimination. In
previous pre-steady-state kinetic studies on the reaction of
Trpase with L-tryptophan using rapid-scanning stopped-flow
spectrophotometry, we found that inclusion of benzimidazole
in reaction mixtures results in the accumulation of a previously
undetected reaction intermediate, with λmax at about 345 nm,
that was assigned to an R-aminoacrylate structure10,14 (Scheme
1). Benzimidazole is an uncompetitive inhibitor (Ki ) 0.81
Scheme 1
Previously, we demonstrated that there is a significant primary
isotope effect of about 3 on the formation of the 345 nm
intermediate when the reaction is performed with R-2H-L-
tryptophan.10 In the present work, we redetermined and con-
firmed the R-deuterium isotope effect of 2.99 ( 0.30 on the
formation of the 345 nm intermediate from R-2H-L-tryptophan
(Figure 2A and Table 1). The amplitude of the absorbance
changes at 345, 422, and 505 nm are also decreased for the
reaction of R-2H-L-tryptophan due to the previously determined
primary isotope effect of 3.6 16 on formation of the quinonoid
mM14) since it is a product analogue of indole that binds only
to the enzyme intermediate after indole release, and hence,
including it makes indole release quasi-irreversible. Furthermore,
(12) Phillips, R. S.; Miles, E. W.; Cohen, L. A. Biochemistry 1984, 23,
6228-6234.
(13) Phillips, R. S.; Miles, E. W.; Cohen, L. A. J. Biol. Chem. 1985,
260, 14665-16470.
(14) Phillips, R. S. Biochemistry 1991, 30, 5927-5934.

1010 J. Am. Chem. Soc., Vol. 122, No. 6, 2000
Phillips et al.
Figure 3. Eyring plot of the temperature dependence of the rate
constant of formation of the 345 nm intermediate.
deuterated), and we found that there is a significant secondary
isotope effect of 1.17 ( 0.03 on the formation of the 345 nm
intermediate (Table 1). R,â,â-Tri-2H-L-tryptophan (55% R,â,â-
trideuterated and 25% R,â dideuterated) was prepared, and we
found that the isotope effects on the reaction are approximately
multiplicative, resulting in the observed isotope effect of 4.42
( 0.67 (Table 1). In addition, we examined the effect of D2O
on the formation of the 345 nm intermediate. There is a
significant solvent isotope effect of 1.79 ( 0.11 on the formation
of the 345 nm intermediate (Table 1), indicating that one or
more protons involved in the transition state are partially
exchangeable with solvent. The solvent isotope effect is also
seen in the reaction of R-2H-L-tryptophan, where the observed
isotope effect is increased from 3 to 4.30 ( 0.16 (Table 1). In
addition, the temperature dependence of the rate constant of
formation of the 345 nm intermediate was examined, and it
exhibits a linear Eyring plot from 5 to 35 °C (Figure 3),
consistent with only a single process contributing to the observed
rate constant. The Eyring activation parameters for the elimina-
tion obtained from Figure 2 are ∆Hq ) 17.2 kcal/mol and ∆Sq
) 6.4 cal/(mol deg).
Isotope Effects on Indole Addition to the r-Aminoacrylate
Intermediate. Addition of L-serine or ammonium pyruvate
together with benzimidazole to Trpase results in a steady-state
spectrum with a peak at 345 nm, very similar to that of the
transient complex seen in the rapid-scanning reactions.14 When
this preformed R-aminoacrylate intermediate is mixed in the
stopped-flow instrument with indole or 5-fluoroindole, a very
rapid reaction is observed that forms a transient quinonoid
intermediate, with λmax at 505 nm. The rate constant for the
formation of the 505 nm peak is dependent on the indole
concentration (Figures 4 and 5). At [indole] > 0.6 mM, the
rate constant becomes so fast that the 505 nm peak is completely
formed within the dead time of the stopped-flow instrument. A
plot of the apparent rate constant for the quinonoid intermediate
formation against [indole] at concentrations below 0.6 mM is
linear, indicating that this is a second-order process reacting
under pseudo-first-order conditions (Figure 5, circles). When
3-2H-indole is used, the rate constant for formation of the
quinonoid intermediate is decreased significantly (Figure 5,
squares). The slopes of the lines fitted by linear regression
analysis to the data in Figure 5 give the apparent second-order
rate constants for the reaction of indole with the R-aminoacrylate
intermediate. These are (1.33 ( 0.11) × 106 M-1 s-1 for indole
and (0.70 ( 0.04) × 106 M-1 s-1 for 3-2H-indole. Thus, there
is a significant isotope effect of 1.88 ( 0.19 on the formation
of the quinonoid intermediate from 3-2H-indole. This second-
Figure 2. (A) Time courses at 345 nm for the reaction of E. coli Trpase
with L-tryptophan and R-2H-L-tryptophan in the presence of benzimid-
azole: solid line, L-tryptophan; dashed line, R-2H-L-tryptophan. (B)
Time courses at 422 nm for the reaction of E. coli Trpase with
L-tryptophan and R-2H-L-tryptophan in the presence of benzimidazole:
solid line, L-tryptophan; dashed line, R-2H-L-tryptophan. (C) Time
courses at 505 nm for the reaction of E. coli Trpase with L-tryptophan
and R-2H-L-tryptophan in the presence of benzimidazole: solid line,
L-tryptophan; dashed line, R-2H-L-tryptophan.
Table 1. Kinetic Isotope Effects on Elimination of Indole from
L-tryptophan
reaction
Dk
R-2H-L-tryptophan in H2O
R,â,â-2H-L-tryptophan in H2O
â,â-2H-L-tryptophan in H2O
L-tryptophan in D2O
2.99 ( 0.30
4.42 ( 0.67
1.17 ( 0.03
1.79 ( 0.11
4.30 ( 0.16
R-2H-L-tryptophan in D2O
intermediate (Figure 2C). We also examined the reaction of â,â-
di-2H-L-tryptophan (47% â,â-dideuterated and 38% â-mono-

Elimination of Indole
J. Am. Chem. Soc., Vol. 122, No. 6, 2000 1011
was in agreement with the intermediacy of the indolenine
tautomer of tryptophan in the reaction mechanisms of both
enzymes.12 Furthermore, the inhibition of Trpase is stereospe-
cific for (3R)-2,3-dihydro-L-tryptophan; hence, we suggested that
the (3R)-indolenine diastereomer of L-tryptophan is the reaction
intermediate.13 In contrast, (3S)-2,3-dihydro-L-tryptophan was
found to be a potent inhibitor of tryptophan synthase.13 Thus,
we proposed that the reaction of indole is a stepwise process in
the mechanism of both Trpase and tryptophan synthase.13 In
the case of Trpase, we proposed a mechanism involving proton
transfer to the indole ring to form a discrete indolenine
intermediate, followed by the carbon-carbon bond cleavage.
In the present work, we performed pre-steady-state kinetic
isotope effect studies in order to obtain kinetic evidence for
the existence of the proposed indolenine intermediate.
Figure 4. Time courses for the reaction of E. coli Trpase with varying
[indole] in the presence of 0.5 M L-serine and 10 mM benzimidazole
at 505 nm. The [indole] are 0.02, 0.04, 0.08, 0.16, and 0.32 mM.
The minimum kinetic mechanism for the Trpase reaction is
given in eq 2, where Eal is the L-tryptophan external aldimine,
K1
\
L-Trp + Trpase y
z Ealyk2z Eqyk3z Eaa + In yk4z Trpase +
k-2
k-3
k-4
-
NH4+ + CH3COCO2 (2)
Eq is the quinonoid intermediate, and Eaa is the aminoacrylate
intermediate. In previous work using single-wavelength and
rapid-scanning stopped-flow methods, we studied the reaction
of L-tryptophan with the enzyme at 25 °C.10,14,15,16 The rate
constant for formation of the external aldimine of L-tryptophan,
Eal, is too fast to measure even by stopped-flow techniques,
but from the concentration dependence of the formation of the
quinonoid intermediate, Eq, we estimated the value of K1 to be
11.6 mM. The deprotonation of the R-carbon of the L-tryptophan
external aldimine is fast, with an extrapolated value of k2 of
about 760 s-1, and the reprotonation rate constant, k-2, is about
60 s-1 10,14,15,16
the formation of the R-aminoacrylate intermediate, with λmax
at 345 nm, occurs with an apparent rate constant, k3, of 32 s-1
.
As we found previously10,14 and in this work,
Figure 5. Concentration dependence of the apparent rate constant for
formation of the 505 nm peak from indole (filled circles) and 3-2H-
indole (filled squares). The lines are the result of linear regression
analysis of the data, with the parameters given in the text.
.
The observed relaxation may be expected to be a combination
of rate constants involving this step and the previous steps.
However, it appears that the R-deprotonation step does not
significantly contribute to the observed rate constant for
formation of the 345 nm intermediate, since there is no
significant concentration dependence of kobs on [L-tryptophan],
whereas the rate constant for formation of the 505 nm quinonoid
intermediate shows a marked dependence on [L-trypto-
phan].10,14,15,16 The clean isosbestic point between the 422 and
505 nm absorbance peaks and the 345 nm intermediate (Figure
1) requires that the external aldimine [Eal] and quinonoid [Eq]
intermediates are at equilibrium during the formation of the 345
nm intermediate. In the presence of benzimidazole, k-3 is
essentially zero in the pre-steady state since the reaction is quasi-
irreversible in the first turnover. Furthermore, the observed rate
constant for formation of the 345 nm intermediate is identical,
within experimental error, to the rate constant, k3, for indole
elimination obtained by fitting rapid chemical quench data to
the proposed mechanism in eq 2 with FITSIM.15 Hence, the
formation of the 345 nm absorbance peak in the presence of
benzimidazole provides a convenient direct spectroscopic
measurement of the first-order rate constant for indole elimina-
tion from the quinonoid intermediate, without significant
complications due to kinetic complexity. Thus, the isotope
effects that we observe on this apparent rate constant are
primarily those on k3, the step of indole elimination from the
quinonoid intermediate of L-tryptophan.
order rate constant is also in good agreement with the kcat/Km
value for indole from steady-state kinetic measurements of 4
× 106 M-1 s-1. The intercepts of the plots in Figure 5 are 78
s-1 for indole and 68 s-1 for 3-2H-indole. For the reaction of
5-fluoroindole, the corresponding rate constants from the slopes
and intercepts of a plot similar to that in Figure 5 (data not
shown) are (6.9 ( 0.1) × 105 M-1 s-1 and 38 ( 2 s-1. This is
also in good agreement with the kcat/Km value of 4 × 105 M-1
s-1 for 5-fluoroindole in steady-state kinetics.
Discussion
The formation of indole during the bacterial putrefaction of
protein was first described by Bopp 150 years ago.4 The
catabolism of L-tryptophan to indole and ammonium pyruvate
by the action of Trpase, an enzyme produced by enteric bacteria,
is now known to be the source of this indole. From a mechanistic
standpoint, the rapid elimination of a carbon leaving group is
intriguing, since it implies that the enzyme has to activate the
indole ring for elimination to take place. Davis and Metzler
proposed in an insightful review published in 197211 that the
indolenine tautomer of tryptophan was an intermediate in the
reaction mechanism of both Trpase and tryptophan synthase,
which catalyzes the synthesis of L-tryptophan from L-serine and
indole. Later, we demonstrated that structural analogues of the
proposed indolenine tautomer, 2,3-dihydro-L-tryptophan and
oxindolyl-L-alanine (Chart 1), are potent competitive inhibitors
of Trpase and tryptophan synthase, and we concluded that this
(15) Lee, M.; Phillips, R. S. BioMed. Chem. 1995, 3, 195-205.
(16) Phillips, R. S. J. Am. Chem. Soc. 1989, 111, 727-730.

1012 J. Am. Chem. Soc., Vol. 122, No. 6, 2000
Phillips et al.
For the mechanism given in eq 2, kcat/Km for L-tryptophan
obtained in steady-state kinetic studies is given by eq 3, since
The reaction of â,â-2H-L-tryptophan shows a normal second-
ary isotope effect on R-aminoacrylate intermediate formation,
as expected for the Câ-C3 bond cleavage, which takes place
with an sp3 to sp2 hybridization change at Câ. The secondary
isotope effect for â,â-dideuterio-L-tryptophan, calculated by
comparison of the reactions of normal and â,â-dideuteriotryp-
tophan, is 1.17. Since the substrate used contained about 65%
deuterium on the â-position, the calculated maximum secondary
isotope effect is 1.28. This is considerably less than the
secondary isotope effect of 1.81 on the formation of the
R-aminoacrylate intermediate from â,â-di-2H-O-acetyl-L-serine
by O-acetylserine sulfhydrylase.19 The reaction of R,â,â-2H-L-
tryptophan also shows a significantly larger isotope effect than
R-2H-L-tryptophan, demonstrating that the primary and second-
ary deuterium isotope effects on the Câ-C3 bond cleavage are
multiplicative. This is consistent with the proton transfer to C3
of indole and the Câ-C3 bond breakage taking place simulta-
neously rather than stepwise. In contrast, for the expected
stepwise mechanism, with proton transfer to the ipso carbon of
the indole to form a discrete indolenine intermediate preceding
Câ-C3 bond cleavage, the observed primary isotope effect
would be expected to decrease for the multiply labeled substrate.
The temperature dependence of the rate constant for R-ami-
noacrylate intermediate formation also shows a linear Eyring
plot (Figure 3), which is consistent with a single step in the
indole elimination. This conclusion is also consistent with
studies of the reaction of 6-(difluoromethyl)tryptophan with
Trpase by Woolridge and Rokita, which showed no evidence
for fluoride elimination and hence argued against deprotonation
of the indole NH in the elimination mechanism.20
kcat/Km ) k2k3/(K1(k-2 + k3))
(3)
indole release is the first irreversible step under initial rate
conditions. Applying the values for the kinetic parameters
obtained in this and previous pre-steady-state studies to eq 3
gives a predicted kcat/Km for L-tryptophan of 3.5 × 104 M-1
s-1, in excellent agreement with steady-state kinetic measure-
ments of 3 × 104 M-1 s-1 17
. Furthermore, in previous steady-
state kinetic isotope effect experiments, we observed a primary
isotope effect on kcat/Km of 2.8 for R-2H-L-tryptophan.17 At that
time, we made a resonable assumption that the only isotope
sensitive step in the reaction of R-2H-L-tryptophan was R-depro-
tonation. However, it is now clear from the present data that
R-deuteration affects both R-deprotonation (k2) and indole
elimination (k3), and thus the observed deuterium isotope effect
on kcat/Km comes from both k2 and k3.
The observed isotope effect of R-2H-L-tryptophan on k3 was
initally surprising. This isotope effect could arise either from
direct transfer with internal return of the R-proton to the indole
leaving group or indirectly through a hydrogen-bonding network.
The intramolecular transfer of a deuterium atom from R-2H-L-
tryptophan to C3 of indole was reported by Vederas et al. in
1978,9 based on NMR analysis of indole isolated from reaction
mixtures of R-2H-L-tryptophan. Depending on the reaction
conditions, the retention of deuterium in the indole product
derived from R-2H-L-tryptophan was only 7.9% in H2O but
100% in D2O. Vederas et al. concluded that the isotope transfer
was intramolecular by examining the product of a mixture of
R-deuterated and ring-deuterated tryptophans.9 Our kinetic
studies are in agreement with this result, as the rate constant
for indole elimination shows a significant isotope effect of about
3 with R-2H-L-tryptophan. An increase in the observed isotope
effect on the elimination to 4.30 is observed when the reaction
of R-2H-L-tryptophan is performed in D2O (Table 1), which
suggests that there is some partial exchange of the R-proton
prior to transfer to the indole ring even in a single turnover.
The value of 4.3 must be the maximum isotope effect on indole
formation, since Vederas et al. found that the indole product
contains 100% deuterium at C3 under these conditions.9 The
solvent and R-substrate deuterium isotope effects on indole
elimination are not strictly multiplicative, indicating that they
are not independent. This would be the case if partial exchange
of a single proton were taking place. A solvent isotope effect
of 1.79 is seen on elimination of indole when the reaction of
L-tryptophan is performed in D2O (Table 1). This result is in
agreement with, but somewhat smaller than, the solvent isotope
effect of 2.5 ( 0.4 that we observed previously on the second
phase of formation of the quinonoid intermediate.16 This
difference likely reflects the experimental difficulty in accurately
fitting the intermediate phase of a three exponential process.
The present experiments isolate the elimination step as a single
exponential process, giving much greater precision in the fitting.
These solvent isotope effects on the reaction of Trpase are
considerably smaller than the solvent isotope effect of 6.9 (
2.3 determined by Kiick from steady-state kinetic measure-
ments.18 However, Kiick determined that multiple waters were
involved and concluded that the solvent effect was due to
changes in solvation of the enzyme. It is possible that a solvent
sensitive conformational change takes place after indole release
and prior to or concomitant with release of the pyruvate product.
The reaction of indole with the preformed R-aminoacrylate
intermediate, prepared by addition to the enzyme of benzimi-
dazole together with L-serine, forms a transient quinonoid
intermediate with λmax at 505 nm (Figure 4), resulting from the
Michael-type addition of indole at C3 to the â-carbon of the
R-aminoacrylate complex. The decay of this transient intermedi-
ate is biphasic, since it has two possible modes of reaction, either
elimination of the indole or reprotonation to tryptophan. For
the reaction of indole with the R-aminoacrylate intermediate,
Eaa, the apparent rate constant for formation of the quinonoid
intermediate, Eq, should be given by eq 4. From the results
1/τ ) k-3[In] + (k-2 + k3)
(4)
shown in Figure 5, which show second-order kinetics under
pseudo-first-order conditions, we can conclude that there is no
kinetically significant noncovalent complex of indole with the
R-aminoacrylate intermediate, Eaa. The value of k-3 of 1.33 ×
106 M-1 s-1 comes from the slope in Figure 5. The intercept of
the plot in Figure 5 should give the sum of the rate constants
of the two ways in which the quinonoid intermediate can break
down, by indole elimination, k3, and R-protonation, k-2. This
intercept value is 76.8 ( 12.3 s-1, in reasonable agreement with
the predicted value of 92 s-1, calculated from 32 s-1 for k3
obtained in the forward direction with benzimidazole and 60
s-1 for k-2 obtained in the reaction of L-tryptophan.10,14,15,16 For
the reaction of 3-2H-indole, the rate constant for quinonoid
intermediate formation, taken from the slope of the line in Figure
4, is 6.4 × 105 M-1 s-1. The isotope effect observed with 3-2H-
indole is thus 1.88, which is consistent with a primary isotope
effect of the C3-indole hydrogen on the formation of the
quinonoid intermediate. A secondary isotope effect might well
(19) Hwang, C. C.; Woehl, E. U.; Minter, D. E.; Dunn, M. F.; Cook, P.
F. Biochemistry 1996, 35, 6358-6365
(20) Woolridge, E. M.; Rokita, S. E. Biochemistry 1991, 30, 1852-1857.
(17) Kiick, D. M.; Phillips, R. S. Biochemistry 1988, 27, 7339-7344.
(18) Kiick, D. M. J. Am. Chem. Soc. 1991, 113, 8499-8504.

Elimination of Indole
J. Am. Chem. Soc., Vol. 122, No. 6, 2000 1013
be expected on the addition of 3-2H-indole to form an indolenine
intermediate. However, such a secondary isotope effect should
be no larger than about 1.2 and, furthermore, should be inverse,
since the hybridization change on C3 in the Michael addition
reaction is from sp2 to sp3. The intercept value for the reaction
of 3-2H-indole, 68 s-1, is in good agreement with the predicted
value of 70 s-1 (10 s-1 for k3 and 60 s-1 for k-2). Hence, in the
direction of tryptophan synthesis, the C3-H bond appears to
be broken concomitant with the formation of the C3-Câ bond.
These results are in contrast to those reported for tryptophan
synthase, where no isotope effect is observed in the reaction of
3-2H-indole with the R-aminoacrylate intermediate.21 Hence,
Lane and Kirschner concluded that the mechanism of indole
reaction for tryptophan synthase is stepwise, with an indolenine
intermediate.21
Thus, the mechanism which is consistent with these results
is given in Scheme 1. External aldimine formation is complete
within the dead time of the stopped-flow instrument, and
subsequent R-deprotonation of the Schiff’s base by a catalytic
base, B1, occurs rapidly to give a quinonoid intermediate with
λmax ) 505 nm. A second base, B2, which interacts with the
indole NH, was identified by steady-state pH dependence
studies.17 Elimination of indole from the quinonoid intermediate
then takes place at 32 s-1, producing an R-aminoacrylate
intermediate with λmax ) 345 nm. This intermediate could be
the R-aminoacrylate Schiff’s base, or it could also be the gem-
diamine adduct with Lys270, as we proposed previously.14 If
the elimination of indole catalyzed by Trpase occurs by a
stepwise mechanism, as we had anticipated, then the reaction
coordinate between the quinonoid intermediate and the R-ami-
noacrylate intermediate should exhibit two maxima and a
relatively deep well for the indolenine intermediate (Chart 2,
In contrast, tryptophan synthase appears to catalyze the synthesis
of tryptophan by the expected stepwise mechanism.21 This
mechanistic difference may explain the very weak activity of
tryptophan synthase to catalyze the elimination of indole from
L-tryptophan.22 There is no apparent chemical precedent for a
concerted SE2 mechanism in similar enzymatic or nonenzymatic
electrophilic aromatic substitution reactions. These consider-
ations suggest that the indolenine structure in the reaction of
Trpase is an intermediate which occupies a broad shallow well
on the reaction coordinate and hence has a very short lifetime.
However, SE2 mechanisms have been proposed for the proto-
nolysis of alkyl mercury compounds by bacterial organomer-
curial lyase,23 and the organomercurial lyase also reacts readily
with aryl mercury compounds. The electrophilic substitution
reaction of indoles at C3 is a very well-known chemical reaction,
and it is generally accepted that indoleninium ions are inter-
mediates in the mechanisms of these reactions.24 However, these
chemical transformations are typically carried out under mod-
erately to strongly acidic reaction conditions. Indole is a rather
weak base, with a pKa for protonation at C3 of about -2,25 and
3-alkyl substituted indoles are even weaker bases than indole,26
so indoleninium ions are extremely unlikely to form under
neutral conditions, even in enzyme active sites. The neutral
indolenine tautomer of indole is also highly unfavorable.25 Since
the range of pKa’s available for general acids in the active site
of enzymes is restricted to values near neutrality, it is possible
that the ipso proton transfer to C3 of tryptophan is so unfavorable
that the steady-state concentration of the indolenine intermediate
in the deep-well model would be extremely small. If the
activation free energies for the reactions leading to and from
the indolenine are reduced to a greater extent than the free
energy of indolenine formation, then the reaction coordinate
with the deep well may readily become converted to that with
the shallow well (Chart 2).
Chart 2
Conclusion. Pre-steady-state kinetic isotope effect studies on
the elimination of indole catalyzed by Trpase have failed to
provide any evidence for the existence of the expected indole-
nine intermediate in the reaction mechanism. Hence, Trpase
apparently uses an unexpected mechanism with very nearly
simultaneous proton transfer and carbon-carbon bond cleavage
to achieve the efficient elimination of indole from L-tryptophan.
left). The deuterated substrate, intermediate, and product are
slightly lower in energy (dotted line). However, if the mecha-
nism were stepwise, then the isotope effects of R- and
â-deuteration should not be multiplicative. Furthermore, for the
deep-well model there should be no primary isotope effect of
deuteration of indole at C3 on the reaction with the R-ami-
noacrylate intermediate. Thus, a stepwise mechanism with a
long-lived indolenine intermediate is not consistent with the data.
The conclusion of these results is that elimination of indole from
the quinonoid intermediate of Trpase is either a concerted
process, with proton transfer to C3 and C3-Câ bond cleavage
ocurring simultaneously (Chart 2, right) or the reaction coor-
dinate has a very broad shallow well for the indolenine species,
so that its lifetime may be limited to only a few bond vibrations
(Chart 2, center). The stereospecificity of inhibition by (3R)-
2,3-dihydro-L-tryptophan13 suggests that this indolenine inter-
mediate or transition state has the (R)-configuration (Scheme
1).
Experimental Section
Materials. L-Tryptophan was purchased from U.S. Biochemical Corp
and was recrystallized from 50% aqueous ethanol before use. The
concentration of L-tryptophan in solutions was determined by the
absorbance at 280 nm (ꢀ ) 5500 M-1 cm-1). Benzimidazole was
purchased from Aldrich Chemical Co. and was recrystallized from hot
water, after treatment with charcoal, before use. S-(o-Nitrophenyl)-L-
cysteine (SOPC) for enzyme assays was prepared from L-cysteine and
2-fluoronitrobenzene as previously described.27 Indole (Gold label) was
obtained from Aldrich, and the concentration of aqueous solutions was
standardized from the absorbance at 270 nm (ꢀ ) 5500 M-1 cm-1).
Deuterated Substrates. R-2H-L-Tryptophan was prepared as previ-
ously described by Kiick and Phillips.17 R,â,â-Tri-2H-L-tryptophan was
(22) Miles, E. W.; Phillips, R. S.; Yeh, H. J. C.; Cohen, L. A.
Biochemistry 1986, 25, 4240-4249.
(23) Begley, T. P.; Walts, A. E.; Walsh, C. T. Biochemistry 1986, 25,
7192-7200.
(24) Sundberg, R. J. Chemistry of Indoles; Academic Press: New York,
1970.
Why would Trpase utilize a concerted or very nearly
concerted mechanism rather than the alternative chemically
reasonable classical stepwise mechanism for indole elimination?
(25) Gut, I. G.; Wirz, J. Angew. Chem., Int. Ed. Engl. 1994, 33, 1153-
1156.
(26) Hinman, R. L.; Lang, J. J. Am. Chem. Soc. 1964, 86, 3796-3806.
(27) Dua, R. K.; Taylor, E. W.; Phillips, R. S. J. Am. Chem. Soc. 1993,
115, 1264-1270.
(21) Lane, A.; Kirschner, K. Eur. J. Biochem. 1983, 129, 571-582.

1014 J. Am. Chem. Soc., Vol. 122, No. 6, 2000
Phillips et al.
prepared by synthesis of L-tryptophan from ammonium pyruvate and
indole catalyzed by Trpase in D2O. Ammonium sulfate (1.50 g) and
potassium pyruvate (1.29 g) were dissolved in 15 mL of D2O (99.8%),
and the stoppered solution was left for 1 week at room temperature.
The solution was then lyophilized, and the resultant yellow solid was
dissolved in 50 mL of D2O. Indole (0.119 g), disodium EDTA (0.0378
g), pyridoxal-5′-phosphate (0.8 mg), and 2-mercaptoethanol (21.8 µL)
were added, and the pD was adjusted to 8.35 with solid K2CO3. Trpase
(16 mg) was then added to the reaction, which was stoppered, covered
with Al foil, and left at room temperature. After 4 days, an additional
16 mg of Trpase was added. After a total of 2 weeks, the reaction
mixture was acidified with acetic acid to pH 4, filtered through Celite,
and extracted with 2 × 20 mL of toluene to remove indole. The aqueous
solution was applied to a column of Dowex 50 (10 × 2.5 cm), washed
with water, and eluted with 1 M aqueous NH3. Evaporation of the
ammonia eluate gave a yellow foam, which was dissolved in 5 mL of
H2O and applied to a column (1 × 20 cm) of reverse phase (C18)
silica gel. The column was washed with 100 mL of H2O and then 100
mL of 10% methanol. The fraction containing the tryptophan was
evaporated to give 7.9 mg of white solid. MS (ESI) showed m/z 206
(no deuterium) (0%), 207 (monodeuterated) (20.2%), 208 (dideuterated)
Enzyme and Assays. Tryptophan indole-lyase was purified from
cells of E. coli JM101 containing plasmid pMD6, with the E. coli tnaA
gene under natural regulation, as described, except that the Sepharose
treatment was performed in a column rather than batchwise.29 Routine
activity assays were performed with SOPC in 0.1 M potassium
phosphate, pH 8.0, at 25° C, following the absorbance decrease at 370
nm (∆ꢀ ) -1860 M-1 cm-1).8 Enzyme concentrations were estimated
from the absorbance of the holoenzyme at 278 nm (A1% ) 9.19),29
using a subunit molecular weight of 52 kDa.30
Kinetic Measurements. Single-wavelength stopped-flow kinetic
measurements and rapid-scanning experiments were performed using
an RSM spectrophotometer and a stopped-flow compartment with a
10 mm path length observation cell from OLIS, Inc. This instrument
has a mixing dead time of about 2 ms; thus, a reaction with a rate
constant of 350 s-1 will lose about half its amplitude during mixing.
Rapid-scanning measurements were performed over the range from 325
to 550 nm at 1 kHz, while single-wavelength measurements for the
concentration dependencies were collected at 4 kHz. Some of the single-
wavelength measurements (Figure 4) were performed on a Kinetics
Instruments stopped-flow mixer with a modified Cary 14 UV/vis
spectrophotometer (OLIS), and preliminary rapid-scanning experiments
were performed with a diode array detector from EG&G Princeton
Applied Research, as previously described.13 Prior to performing the
rapid kinetics experiments, the stock enzyme was incubated with 0.5
mM PLP for 1 h at 37 °C and then separated from excess PLP on a
short desalting column (PD-10, Pharmacia) equilibrated with 0.02 M
potassium phosphate, pH 8.0, and 0.16 M KCl, and the reactions were
performed in the same buffer. For stopped-flow solvent isotope effect
studies, the enzyme was concentrated in an Amicon cell to about 0.1
mL, diluted with 2 mL of buffer in D2O, concentrated to 0.1 mL, diluted
again with 2 mL of buffer in D2O, concentrated again to 0.1 mL, and
then diluted with 1 mL of buffer in D2O. The stopped-flow kinetic
measurements were performed at 25 °C with the stopped-flow compart-
ment thermostated by an external water bath. Generally, the enzyme
solutions, in buffer with 10 mM benzimidazole, were mixed with 20
mM L-tryptophan or deuterated tryptophan in the stopped-flow instru-
ment. In the reverse reaction, enzyme solutions in 0.5 M L-serine with
10 mM benzimidazole were mixed with solutions of indole in the same
buffer. Time courses at selected wavelengths were analyzed by fitting
with the SIFIT or LMFIT programs (eq 5) (OLIS, Inc.), which can fit
1
(26.5%), and 209 (trideuterated) (53.3%). H NMR indicated that the
R-H was >99% exchanged.
â,â-Dideuterio-L-tryptophan was prepared by the reaction of indole
with R,â,â-trideuterio-S-ethyl-L-cysteine catalyzed by Trpase in water.
R,â,â-Trideuterio-S-ethyl-L-cysteine was prepared from normal S-ethyl-
L-cysteine by an isotope exchange reaction in D2O catalyzed by
methionine-γ-lyase. The lyophilized cells of Citrobacter intermedius
containing methionine-γ-lyase were obtained as described28 and used
as a catalyst. Lyophilized cells (100 mg) were added to the solution of
150 mg of S-ethyl-L-cysteine in 0.1 M potassium phosphate buffer in
D2O, pD 8.8, in the presence of 0.1 mM PLP, and the mixture was
incubated at 30 °C with stirring for 3 days. The cells were separated
by centrifugation, and the solution was analyzed by NMR, which
showed that, together with the partial decomposition of the starting
S-ethyl-L-cysteine, extensive isotope exchange of its R,â,â protons
occurred. The amino acid was isolated on a Dowex-50 column, and
about 70 mg of a brown material was obtained. A portion (30 mg) of
this material was dissolved in 10 mL of 0.1 M potassium phosphate
buffer, pH 8.5, containing 0.1 mM PLP and 15 mg of indole, and 0.05
mL of Trpase solution (0.82 mg) was added. The mixture was incubated
with stirring at room remperature for 24 h. The enzyme was denatured
by heating at 100 °C for 5 min and removed by centrifugation. The
unreacted indole was extracted with ether, and amino acids were isolated
from the solution on a Dowex-50 column. The â,â-dideuterio-L-
tryptophan was purified on a reverse phase (C 18) silica gel column as
described above. Its identity was proven by TLC (single spot on a silica
gel plate in n-butanol-acetic acid-water (4:1:1)) and NMR analysis.
MS (ESI) analysis revealed the presence of 47% dideuterated, 38%
monodeuterated, and 15% nondeuterated tryptophan. 1H NMR analysis
showed that all of the deuterium was on the â-carbon.
1-3
aie-k t + c
(5)
i
At )
∑
i
up to three exponentials with amplitudes and an offset, where At is the
absorbance at time t, ai is the amplitude of each phase, ki is the rate
constant for each phase, and c is the final absorbance, if nonzero.
Quality of fit was judged by analysis of the residuals and by the
Durbin-Watson value.31 The concentration dependencies of relaxations
in Figure 5 were fit by linear regression to obtain the second-order
rate constants. The reactions were replicated from 3 to 14 times, and
the mean rate constants were determined. The standard deviations were
generally 10% or less of the parameter mean values. The mean values
were then used to calculate the isotope effects.
3-2H-Indole was prepared by the reaction of Trpase in D2O.
L-Tryptophan (0.200 g), K2HPO4 (0.313 g), KH2PO4, pyridoxal-5′-
phosphate (0.25 mg), and 20 mL of D2O were combined, and 0.5 mL
of Trpase (5 mg) was added. The flask was sealed with a serum stopped
and incubated in a shaker bath at 37 °C for 2 days. The reaction mixture
was then extracted with 2 20 mL portions of CH2Cl2. The extracts were
dried over Na2SO4 and evaporated to give 14 mg of indole, as light
yellow crystals. 1H NMR of the indole indicated that it contained 93%
2H at C3.
Acknowledgment. This work was partially supported by a
grant from NIH (GM42588) to R.S.P.
JA991647Q
(29) Phillips, R. S.; Gollnick, P. D. J. Biol. Chem. 1989, 264, 10627-
10632.
(28) Faleev, N. G.; Troitskaya, M. V.; Paskonova, E. A.; Saporovskaya,
M. B.; Belikov, V. M. Enzymol. Microb. Technol. 1996, 19, 590-593.
(30) Deeley, M. D.; Yanofsky, C. J. Bacteriol. 1981, 147, 787-796.
(31) Durbin, J.; Watson, G. S. Biometrika 1970, 37, 409-414.