Mechanism for the direct synthesis of tryptophan from indole and serine: A useful NMR technique for the detection of a reactive intermediate in the reaction mixture

Article abstract of DOI:10.1002/mrc.2676

The reaction mechanism for the biomimetic synthesis of tryptophan from indole and serine in the presence of Ac₂O in AcOH was investigated. Although the time-course ¹H-NMR spectra of the reaction of 5-methoxyindole with N-acetylserine were measured in the presence of (CD ₃CO)₂O in CD₃CO₂D, the reactive intermediate could not be detected. This reaction was conducted without 5-methoxyindole in order to elucidate the reactive intermediate, but the intermediate could not be isolated from the reaction mixture. Since the intermediate would be expected to have a very short life time, and therefore be very difficult to detect by conventional analytical methods, the structure of the intermediate was elucidated using a 2D-NMR technique, diffusion-ordered spectroscopy (DOSY). Two intermediates were detected and confirmed to be 2-methyl-4-methyleneoxazol-5(4H)-one and 2-methyl-4-hydroxymethyloxazol-5(4H)- one. The present results demonstrated that DOSY is a powerful tool for the detection of unstable intermediates.

Full text of DOI:10.1002/mrc.2676

Research Article
Received: 14 June 2010
Revised: 21 July 2010
Accepted: 23 July 2010
Published online in Wiley Online Library: 26 August 2010
(wileyonlinelibrary.com) DOI 10.1002/mrc.2676
Mechanism for the direct synthesis
of tryptophan from indole and serine: a useful
NMR technique for the detection of a reactive
intermediate in the reaction mixture
Yuusaku Yokoyama,a∗ Masamichi Nakakoshi,a Hiroaki Okuno,a
Yohko Sakamotoa and Satoshi Sakuraib
The reaction mechanism for the biomimetic synthesis of tryptophan from indole and serine in the presence of Ac2O in AcOH was
investigated. Although the time-course 1H-NMR spectra of the reaction of 5-methoxyindole with N-acetylserine were measured
in the presence of (CD3CO)2O in CD3CO2D, the reactive intermediate could not be detected. This reaction was conducted
without 5-methoxyindole in order to elucidate the reactive intermediate, but the intermediate could not be isolated from
the reaction mixture. Since the intermediate would be expected to have a very short life time, and therefore be very difficult
to detect by conventional analytical methods, the structure of the intermediate was elucidated using a 2D-NMR technique,
diffusion-ordered spectroscopy (DOSY). Two intermediates were detected and confirmed to be 2-methyl-4-methyleneoxazol-
5(4H)-one and 2-methyl-4-hydroxymethyloxazol-5(4H)-one. The present results demonstrated that DOSY is a powerful tool for
c
the detection of unstable intermediates. Copyright ꢀ 2010 John Wiley & Sons, Ltd.
Supporting information may be found in the online version of this article.
Keywords: NMR; tryptophan; synthesis; indole; serine; biosynthesis; mechanism; DOSY
Introduction
We previously proposed[2] a plausible mechanism, shown
in Scheme 2. An α,β-unsaturated azlactone, 2-methyl-4-
methyleneoxazol-5(4H)-one (9) is a key intermediate which
acts as a reactive Michael acceptor towards 4-bomroindole
(1). The intermediate (9) formed from serine azlactone (7) or
2-acetamideacrylate (8) via N-acetylserine (6). It was reported[5]
thatthereactionof 8withindolegave N-acetyltryptophanein55%
yield. In contrast, Sanderson proposed[4] another intermediate,
cyclic oxonium ion (11), which is attacked by indole in an SN2-type
reaction.
Herein we report a detailed investigation of the mechanism of
this reaction using a new 2D-NMR technique, diffusion-ordered
spectroscopy (DOSY). DOSY is a useful method for obtaining
structural information on unstable intermediates in the reaction
mixture.
Tryptophan and its analogs have interesting biological activities.
Numerous synthetic methods for obtaining racemic and optically
active tryptophan derivatives have been reported, and the
development of efficient synthetic methods is of interest to many
synthetic chemists. Of these methods, biomimetic-type syntheses
are very attractive because one-step synthesis from serine and
indole is not only practical but also mechanistically interesting.
In this approach, an SN2-type reaction occurs directly at the C3-
position of the indole with an inactivated hydroxyl group attached
at the carbon atom of L-serine. Although there have been several
reports[1] of the direct synthesis of tryptophan from indole and
serine or their analogs, the yields were unsatisfactory (3%), so this
approach is not practical.
During the course of our investigation of the total synthesis of
ergot alkaloids, we developed[2] a two-step synthesis of optically
pure (S)-4-bromotryptophan (4) from 4-bromoindole (1) and dl-
serine(2)involvingthekineticresolutionofacylase(Scheme1).The
one-step formation of N-acetyl-4-bromotryptophan (3) was the
first example of the practical biomimetic synthesis of tryptophan.
Yamada’s group[3] applied this method to the synthesis of
optically pure 7-bromotryptophan. Recently, Sanderson et al. also
reported[4] the synthesis of optically active tryptophan analogs
having various substituents on the indole ring. However, the
drawback of this reaction is the formation of racemers even if
optically active L-serine is used. Therefore, in order to develop
an asymmetric synthesis strategy, a precise understanding of the
mechanism is indispensable.
Results and Discussion
The reactions shown in Scheme 2 proceed through N-acetylserine
(6) as an initial intermediate. We selected the reaction of
∗
Correspondence to: Yuusaku Yokoyama, Faculty of Pharmaceutical Sciences,
Toho University, 2-2-1 Miyama, Funabashi, Chiba 274-8510, Japan.
E-mail: yokoyama@phar.toho-u.ac.jp
a
FacultyofPharmaceuticalSciences,TohoUniversity,2-2-1Miyama,Funabashi,
Chiba 274-8510, Japan
b
JEOL Ltd, Musashino, 3-1-2 Akishima, Tokyo 196-8556, Japan
c
Magn. Reson. Chem. 2010, 48, 811–817
Copyright ꢀ 2010 John Wiley & Sons, Ltd.

Y. Yokoyama et al.
Scheme 1. Two-step synthesis of optically active 4-bromotryptophan.
Scheme 2. Plausible mechanism for the formation of N-acetyltryptophan (3).
5-methoxytryptophan methyl ester (13a), in 79% yield after
esterification with (CH3)3SiCHN2 (Scheme 3).
Thetime-course1H-NMRspectraofthisreactionweremeasured
in the presence of (CD3CO)2O in CD3CO2D. The spectra are shown
inFig. 1. Asthereactionproceeds, thepeakdueto12atδ 7.15 ppm
(peak a) disappears, while the peaks due to the methylene proton
of 13b (peaks b) become prominent (Fig. 1A–D). Although a pair
of double doublets (peaks c, δ 3.8 and 4.1 ppm) arising from
the methylene proton of 6 remain after 15 min, by which time
starting material 12 has almost disappeared (Fig. 1C), those peaks
disappear after 60 min, and unknown new peaks at δ 6 and
δ 6.4 ppm (peaks d) appear (Fig. 1D).
Scheme 3. Reaction of 5-methoxyindole (12) with N-acetylserine (6).
Since these new peaks should be due to the reaction of 6
with Ac2O, the reaction of 6 with Ac2O-d6 was carried out and
NMR spectra were measured (Fig. 2A–C). Although the same two
sets of peaks (peaks d) appeared in the reaction mixture, these
6 with 5-methoxyindole (12) as a model reaction for easy
assignment of the aromatic part of the 1H-NMR spectrum. The
reaction proceeded smoothly to give the product, N-acetyl-
c
wileyonlinelibrary.com/journal/mrc
Copyright ꢀ 2010 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2010, 48, 811–817

Direct synthesis of tryptophan from indole and serine
Figure 1. 1H-NMR spectrum of the reaction of 5-methoxyindole (12) with 2.0 equiv. of dl-serine (6) at 80 ◦C in the presence of (CD3CO)2O in CD3CO2D,
obtained using a 400-MHz NMR. (A) 0 min, (B) 5 min, (C) 15 min, (D) 60 min and (E) 5-methoxytryptophan (13b) with Ac2O-d6 in CD3CO2D at 80 ◦C after
60 min.
Figure 2. 1H-NMR spectra of the reaction of N-acetylserine (6) (A–C) and 2-acetamidoacrylic acid (8) (D and E) with (CD3CO)2O in CD3CO2H at 60 ◦C,
obtained using a JEOL ECP-400 (400 MHz).
peaks were assigned to the olefinic protons of 2-acetamidoacrylic
acid (8) (Fig. 2E), which is known[6] to form under these reaction
conditions. We also observed the formation of 8 by heating of
serine (2) or N-acetylserine (6) with Ac2O in AcOH followed by
direct evaporation of the solvent and Ac2O. However, two new
sets of peaks were observed at δ 4.5 ppm (peak e) and δ 6.0 ppm
(peaks f) during the reaction of 6 with 12 (Fig. 2B and C). Peaks f
were also observed in the reaction of 8 with Ac2O-d6 after 60 min
(Fig. 2D). We speculated that peaks e and f arose from the reactive
intermediates.
based on the different diffusion coefficients of compounds. The
DOSY spectrum and the slice data are shown in Figs 3 and 4,
respectively. The slice spectrum in Fig. 4A shows the presence of
an unstable intermediate and the slice spectrum in Fig. 4C shows
a mixture of 6 and another unstable intermediate. These unknown
intermediates were speculated to be oxazolone 7 and 9, derived
from 6 (Scheme 1), as shown in Fig. 4A and C.
The structures of these intermediates were further confirmed
on the basis of various 2D NMR spectra. The 1H- and 13C spectra
of the reaction mixture after 60 min are shown in expanded form,
the range from δ 3.5 to 6.6 ppm, in Fig. 5. In the 1H-detected pulse
fieldgradientmultiplequantumcoherence(PFG-HMQC)spectrum
The above speculation was strongly supported by the results of
DOSY, which is a useful 2D NMR method for spectrum separation[7]
c
Magn. Reson. Chem. 2010, 48, 811–817
Copyright ꢀ 2010 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc

Y. Yokoyama et al.
Figure 3. 1H-DOSY spectrum of the reaction of N-acetylserine with Ac2O-d6 obtained using a JEOL ECX-400 (400 MHz).
Figure 4. Slice spectra of Fig. 3.
1
for H–13C correlations, cross peaks were observed between Hc
and Cc for compound 7 and Hg and Cg for compound 9, and
in the detected pulse field gradient multiple-bond heteronuclear
multiple bond coherence (PFG-HMBC) spectrum, cross peaks were
observed between Hc and Cd for compound 7 and Hg and Ch for
compound 9. [The data are shown in the supporting information.]
All of the peaks for compounds 6, 7, 8 and 9 could be rationally
assigned, as shown in Fig. 5.
derived from N,O-diacetylserine (14) is the key intermediate which
is reactive in nucleophilic substitution reactions, and that 11
equilibrates to N-acetyldehydroalanine (8). They also mentioned
that N, O-diacetylserine (14) was obtained as a major product from
the reaction of 2 and Ac2O, without the formation of 8. In contrast,
in our NMR experiments, there is no evidence for the formation
of 14 or intermediate 11: there are no signals other than those
due to 6, 7, 8 and 9 in the reaction mixture of 6 with Ac2O-d6
in AcOH-d4, even after 60 min, as shown in Fig. 2C. Furthermore,
the olefinic protons of 8 gradually increased in intensity with time
(Fig. 2A–2C), and the broad triplet at δ 4.7 ppm, which was the
asymmetric proton of N-acetylserine (6), was gradually decreased
by the exchange of deuterium proton through oxazolone (7). This
result clearly indicates that 8 is a major by-product and that N,
O-diacetylserine (14) is not formed in the reaction mixture.
In addition, 1H–15N PFG-HMBC in Fig. 6 strongly supported the
above estimation, i.e. the spectrum showed cross peaks between
15N signals at δ 370 and 495 ppm and 1H methylene signals of 7
and olefinic protons of 9, in addition to the cross peaks of 6 and 9.
Although all spectral data showed that our proposed
mechanism is rational, Sanderson[4] proposed a different mech-
anism (Scheme 4). They concluded that the cationic species (11)
c
wileyonlinelibrary.com/journal/mrc
Copyright ꢀ 2010 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2010, 48, 811–817

Direct synthesis of tryptophan from indole and serine
Figure 5. Assignment of 1H- and 13C-NMR spectra based on two-dimensional spectra (H–H DQFCOSY, H–C PFG-HMQC and PFG-HMBC) obtained using
a JEOL ECP-500 (500 MHz).
Figure 6. 1H–15N HMBC spectrum. Reference for 15N is CH3C15N at 249 ppm. Obtained using a JEOL ECP-500 (500 MHz).
Conclusion
anhydride, we confirmed that the intermediate is impossible to
isolate. The present results shows that NMR measurement, espe-
cially DOSY, is a powerful tool for the detection and structure
elucidation of unstable intermediates in the reaction mixture. The
Using
a new 2D-NMR technique, DOSY and other NMR
measurement techniques for the reaction of serine with acetic
c
Magn. Reson. Chem. 2010, 48, 811–817
Copyright ꢀ 2010 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc

Y. Yokoyama et al.
Scheme 4. Proposed mechanism by Sanderson[4]
.
asymmetric Michael addition might be a guiding principle for the
present reaction in order to obtain optically active tryptophan
derivatives. We are now applying this reaction to asymmetric
syntheses based on this mechanistic principle.
NMR experiment for the reaction of N-acetylserine with
Ac2O-d6
A mixture of N-acetylserine (dl-6) (62 mg, 0.43 mmol) and Ac2O-d6
(0.127 ml, 138 mg, 0.43 mmol) in Ac2O-d6 (0.7 ml) was heated in
an NMR tube at 80 ◦C for 1.5 h under argon. 1H and 13C NMR
spectra were measured on a JEOL ECP-500 NMR equipped with
TH5ATFG2 probe at 27◦. The diffusion times were optimized for
every experiment. The gradient strengths used in this study were
between 3[T/M] − 0.3[T/M], and the 90◦ pulse of 1H is 12.7 µs.
1H-observed field gradient (FG) double-quantum filtered
correlation spectroscopy (FG-DQFCOSY) was used for the mea-
surement of absolute values. The double pulsed-field-gradient-
spin-echo (DPFGSE) sequence was used as the excitation sculpting
technique.[8] The effect was to selectively invert protons attached
to carbon-12 while leaving intact protons bound to carbon-13
nuclei. For the DPFGSE experiments a recycle delay of 2 s was
employed, operating at 500 MHz and 80 ◦C using a deuterium
lock system. Filled rectangles are 90◦ pulses; open rectangles are
180◦ pulses that were conventional 180◦ pulses on the proton
channel and composite 90◦x, 240◦y, 90◦x inversion pulses on the
13C channel.
Experimental
Synthesis of 5-methoxytryptophan metyl ester (13b)
with N-acetylserine (6)
A mixture of 12 (150 mg, 1.0 mmol), dl-N-acetylserine (dl-6)
(308 mg, 2.1 mmol), and Ac2O (0.4 ml, 4.0 mmol) in AcOH (3.6 ml)
was heated at 80 ◦C for 1.5 h under argon. The mixture was
basified with 30% KOH in an ice bath and washed with organic
solvent (AcOEt : benzene = 1 : 1). The aqueous layer was acidified
with conc. HCl and extracted 3 times with AcOEt. The combined
organic layers were washed with H2O and brine and dried over
MgSO4. After evaporation of solvent and azeotropic removal of
AcOH with benzene, the resulting brown solid (337 mg) was
dissolved in AcOEt (4 ml) and MeOH (2 ml). To this solution,
2.0 M trimethylsilyldiazomethane (TMSCHN2, 2.7 ml, 5.5 mmol) in
hexane was added and kept for 30 min at room temperature.
Then, AcOH was added to quench the reaction and the solution
was diluted with water. The aqueous layer was extracted 3 times
with AcOEt and the combined organic layers were washed with
saturated aqueous NaHCO3 and brine and dried over MgSO4. After
solvent evaporation, the resulting brown oil was subjected to
silica gel chromatography (4 : 1 benzene : acetone) to give 231 mg
of 13b (79% yield) as a colorless amorphous. 1H NMR (CDCl3,
400 MHz) δ 1.97 (s, 3H), 3.29 (m, 2H), 3.71 (s, 3H), 3.85 (s, 3H), 4.95
(m, 1H), 5.99 (br d, J = 8.0 Hz, 1H), 6.85 (dd, J = 8.8, 2.4 Hz, 1H),
6.95 (d, J = 2.4 Hz, 1H), 6.98 (d, J = 2.4 Hz, 1H), 7.25 (d, J = 8.8 Hz,
1H), 8.01 (br s, 1H). IR νmax (KBr) 3404, 1739, 1655 cm−1 EI–MS
m/z 290 (M+, 70%), 160 (base peak). Anal. calcd for C15H18N2O4
(290.31) : C, 62.06; H, 6.25; N, 9.95. Found: C, 61.99; H, 6.53; N, 9.39.
The BIRD elements are labeled by the phase of the central
proton 180◦ pulse. All FG pulses were 1 ms followed by a recovery
delay of 200 µs.
On the basis of the 13C–1H correlations obtained from
the DPFGSE HMQC spectra, we assigned the 13C chemical
shifts corresponding to the 1H chemical shifts. A long-range
heteronuclearcorrelationwasobservedinthe1Hspectraobtained
by the FG-HMBC measurements. Quaternary carbon atoms were
assigned on the basis of the long-range correlation.
Two-dimensional 1H-observed DPFGSE HMQC and HMBC
spectra permitted distinction between 1H and 13C (1JH–C) by
restricting the observational frequency range.
1H–15N HMBC spectra can lead to significantly better results as
shown in Fig. 6. CH3C15N at 249 ppm was used as a reference, and
the spectra were obtained on a JEOL ECP-500.
NMR experiment for the reaction of 5-methoxyindole
with N-acetylserine
DOSY measurements
A mixture of N-acetylserine (dl-6) (43 mg, 0.29 mmol) and Ac2O-d6
(0.088 ml, 96 mg, 0.30 mmol) in d6-AcOH (0.7 ml) was heated in
an NMR tube at 80 C for the times indicated in Fig. 1A–D. H-
NMR spectra were measured at 80 ◦C on a JEOL JNM-ECP-400 NMR
spectrometerandwererecordedinδ units, partspermillion(ppm).
The chemical shifts were measured relative to tetramethylsilane.
1H-DOSY[9] spectra were measured on a JEOL JNM-ECA-500
equipped with a TH5FG probe-head with an actively shielded
Z-gradient coil. 1H-DOSY used the bipolar pulse pairs stimulated
echo-longitudinal eddy current delay (BPPSTE-LED)[10] pulse
sequence at 30 ◦C. The gradient amplitude (g) was changed
from 1 to 30 G cm−1 in 32 steps. Measurement conditions for
◦
1
c
wileyonlinelibrary.com/journal/mrc
Copyright ꢀ 2010 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2010, 48, 811–817

Direct synthesis of tryptophan from indole and serine
1H-DOSY were diffusion time 0.3 s, FG pulse width 1.2 ms and
eddy current delay 50 ms. We used the SPLMOD algorithm for
DOSY processing supported by JEOL (DELTA version 4.34). In the
supporting information, we showed the DOSY spectra and slice
data which were treated with Levenberg-Marquardt method as
one of the most reliable DOSY algolism.
J. Am. Chem. Soc. 1985, 107, 4093; (c) Y. Murakami, Y. Hisaeda,
T. Miyajima, H. Sakata, J. Kikuchi, Chem. Lett. 1993, 645, and
references cited therein.
[2] (a) Y. Yokoyama, H. Hikawa, M. Mitsuhashi, A. Uyama, Y. Murakami,
Tetrahedron Lett. 1999, 40, 7803; (b) Y. Yokoyama, M. Hikawa,
M. Mitsuhashi, A. Uyama, Y. Hiroki, M. Murakami, Eur. J. Org. Chem.
2004, 1244.
[3] Y. Konda-Yamada, C. Okada, K. Yoshida, Y. Umeda, S. Arima, N. Sato,
T. Kai, H. Takayanagi, Y. Harigaya, Tetrahedron 2002, 58, 7851.
[4] G. Blaser, J. M. Sanderson, A. S. Batsanov, J. A. K. Howard, Tetrahe-
dron Lett. 2008, 49, 2795.
Acknowledgements
This work was supported by a grant-in-aid for Scientific Research
(20590014) from the Ministry of Education, Science, Sports and
Culture, Japan (MEXT), and by ‘Open Research Center’ project for
private university matching fund subsidy from MEXT, and also
supported by a Uehara Memorial Foundation.
[5] H. R. Snyder, J. A. MacDonald, J. Am. Chem. Soc. 1955, 77, 1257.
[6] M. Nakagawa,Y. Torisawa,T. Hosoka,K. Tanabe,F. Tavet,M. Aikawa,
T. Hino, Heterocycles 1993, 35, 1167.
[7] M. Nakakoshi, M. Ueda, S. Sakurai, O. Miyata, M. Sugiura, T. Naito,
Magn. Reson. Chem. 2006, 44, 807.
[8] J. L. Marshall, Carbon-Carbon and Carbon-Proton NMR Couplings:
Application to Organic Stereochemistry and Conformational Analysis,
Verlag Chemie International: Deerfield Beach, 1983, p 22; (b) T.-
L. Hwang, A. J. Shaka, J. Magn. Reson. 1995, A112, 275.
[9] (a) K. F. Morris, C. S. Johonson, J. Am. Chem. Soc. 1992, 114, 3139;
(b) K. F. Morris, C. S. Johonson, J. Am. Chem. Soc. 1993, 115, 4291.
[10] D. Wu., A. D. Chen, C. S. Johnson Jr, J. Magn. Res., Series A 1995, 115,
260.
Supporting information
Supporting information may be found in the online version of this
article.
References
[1] (a) D. E. Metzler, M. Ikawa, E. E. Snell, J. Am. Chem. Soc. 1954, 76, 648;
(b) W. Weiner, J. Winkler, S. C. Zimmerman, W. Czarnik, R. Breslow,
c
Magn. Reson. Chem. 2010, 48, 811–817
Copyright ꢀ 2010 John Wiley & Sons, Ltd.
wileyonlinelibrary.com/journal/mrc