Chemical development of MDL 103371: An N-methyl-D-aspartate-type glycine receptor antagonist for the treatment of stroke

Article abstract of DOI:10.1021/op000286s

MDL 103371 is a N-methyl-D-aspartate (NMDA)-type glycine receptor antagonist for the potential treatment of stroke. Evaluation of five different synthetic routes, which included Stille, Suzuki, enol ether, Knoevenagel, and the Mukaiyama coupling reactions

Full text of DOI:10.1021/op000286s

Organic Process Research & Development 2000, 4, 477−487
Chemical Development of MDL 103371: An N-Methyl-D-Aspartate-Type Glycine
Receptor Antagonist for the Treatment of Stroke†
Timothy J. N. Watson,* Stephen W. Horgan, Ramnik S. Shah, Robert A. Farr, Richard A. Schnettler,
C. Richard Nevill, Jr., Franz J. Weiberth, Edward W. Huber, Bruce M. Baron, Mark E. Webster,‡ Rajesh K. Mishra,‡
Boyd L. Harrison,§ Phillip L. Nyce,| Cynthia L. Rand, and Christian T. Goralski
AVentis Pharmaceuticals, Chemical DeVelopment, 2110 East Galbraith Road, Building 51, Cincinnati, Ohio 45215-6300,
U.S.A.
Abstract:
utilized by glutamate. Thus, glycine antagonists are non-
competitive antagonists of glutamate and offer a potential
mechanistic advantage since they do not have to compete
with the ischemia-induced elevated levels of glutamate. We
have demonstrated that glycine antagonists can antagonize
NMDA receptor-mediated responses in vitro and in vivo.
Therefore, glycine antagonists offer a novel mechanism to
provide neuroprotection in stroke and trauma.
One such compound, MDL 103371 (1), was selected as
a potential drug candidate for the treatment of stroke. It
became necessary to develop a synthesis of 1 that would
provide the multikilogram quantities needed for further
development.
MDL 103371 is a N-methyl-D-aspartate (NMDA)-type glycine
receptor antagonist for the potential treatment of stroke.
Evaluation of five different synthetic routes, which included
Stille, Suzuki, enol ether, Knoevenagel, and the Mukaiyama
coupling reactions, revealed the Knoevenagel approach superior
for preparing large quantities of drug substance for evaluation.
The overall process utilized some classical chemistry. Fischer
indole cyclization, followed by a Vilsmeier-Haack formylation
and a Knoevenagel condensation gave immediate access into
the proper carbon framework of the target molecule. A unique
hydrogenation cataylst and solvent system for a nitro reduction,
followed by a two step acid-base hydrolysis of a nitrile gave
the crude product. Purification was accomplished by a potas-
sium salt crystallization followed by a Schiff base formation to
give MDL 103371 in nine steps in an overall yield of 38%.
Introduction:
Brain ischemia or concussive injury is accompanied by a
dramatic rise in the extracellular concentrations of glutamate
and aspartate. These amino acids function as neurotransmitter
substances and are capable of activating an ion channel-
linked receptor with a high permeability to calcium ions,
the N-methyl-D-aspartate (NMDA)-type glutamate receptor.
A variety of experimental observations link excessive or
prolonged glutamate receptor stimulation with cell death
occurring following ischemia or traumatic injury. Glutamate
antagonists have been shown to be of benefit in animal
models of stroke and trauma by greatly increasing neuronal
survival, reducing animal mortality, and improving neuro-
logical outcome.
One can envision MDL 103371 (1) as a 3-aminophenyl-
acetic acid linked to a dichloro-indole acid through an
E-olefin, thus placing the diacid functionality in a syn spatial
relationship. The FDA’s policy1 in developing geometric
isomers states, “Geometric isomers and diastereomers there-
fore should, with the exception of cases where interconver-
sion occurs, be treated as separate drugs and developed
accordingly.” In fact, in vivo interconversion from the E to
Z isomer does not take place, and the Z isomer was
determined to be inactive. Second, MDL 103371 is sensitive
to isomerization in the presence of light and in a variety of
organic and aqueous solutions. However, to our benefit, the
E isomer is generally favored, depending on the conditions.
Along with developing a synthesis that could prepare
kilogram quantities of 1, we had to address the isomerization
and stability issues and develop a purification method to
isolate the E isomer exclusively.
Glycine has been shown to be an obligatory co-agonist
for the NMDA receptor activating at a site distinct from that
† This paper was presented at the Hope Symposium, Hope College, June 11,
1999, and the Organic Reactions & Processes Gordon Conference July 11-16,
1999, by Timothy J. N. Watson. This paper will follow the outline of the lecture
presented and only incorporate the experimentals from the scale-up work.
* To whom correspondence should be addressed. Current Address: Pfizer
Process Research & Development, Pfizer Inc., Eastern Point Road, Groton, CT
06340. E-mail: timothy•j•watson@groton.pfizer.com.
‡ Current address: Procter & Gamble, 8700 Mason-Montgomery Road,
Mason, OH 45040-9462.
Route Selection
§ Current Address: Associate Director, Chemical Sciences Wyeth-Ayerst
Research CN, 8000 Princeton, NJ 08543.
Potential routes will be discussed in general terms with
an emphasis on olefin formation to form the proper carbon
framework of the desired system and issues pertaining to
| Current Address: Vertex Pharmaceuticals Inc., 130 Waverly Street,
Cambridge, MA 02139.
Contract Manufacturing Services R&D, The Dow Chemical Company, 1710
Building, Midland, MI 48674.
(1) www.fda.gov/cder/guidance/stereo.htm.
10.1021/op000286s CCC: $19.00 © 2000 American Chemical Society and The Royal Society of Chemistry
Published on Web 09/20/2000
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
477

Scheme 1
The Suzuki approach was attractive to prepare a wide
variety of aryl analogues from a single late stage intermedi-
ate. The ability to differentiate between the two ester
functionalities added diversity to the analogues. However,
from a scale-up perspective it added two steps with a
protection and deprotection of the indole. The complex
reaction mixtures required chromatography and generally
took long times for the reaction to reach completion.
Anhydrous Suzuki conditions had to be developed because
of the labile nature of the tosylate under standard conditions.
This approach received a low priority. Any attempts to
directly prepare the desired analogue with the appropriate
Horner-Emmons reagent were unsuccessful. Some of the
analogues prepared were the phenyl, p-chlorophenyl, 3,4-
dichlorophenyl, 2- and 3-thiophene and furan, and the
m-methoxyphenyl.
scale-up. The routes discussed were developed for screening
a variety of possible drug candidates where the aniline
portion of the structure was altered; MDL 103371 (1) was
the lead compound generated from this work. Chemical
Development used the various routes to help prioritize our
options. Complete details on the specific analogues will not
be discussed.2,3
The initial attempts to prepare the p-fluoro analogue via
the Suzuki coupling with the vinyl bromide and commercially
available p-fluorophenylboronic acid used method A. How-
ever, because of the complexity of the product mixture
obtained and the difficulty in purification of the product, a
third approach was explored involving a Lewis acid catalyzed
condensation reaction of the indole ester with an appropriate
enol ether (see Scheme 3). A variety of enol ethers were
condensed with the indole to give direct access to the proper
carbon framework. This approach was the route that Chemi-
cal Development utilized to prepare the first sample of MDL
103371 (more specifics of this approach will be discussed
in this section because we did use this work to make a 25 g
sample). Reaction of 3-nitrophenylacetate with sodium
methoxide and methyl formate, followed by MeI gave the
enol ether in modest yield. Upon scaling up this chemistry,
we found out the enol ether would slowly decompose at room
temperature. The decomposition was difficult to handle on
larger scale due to the longer times required to isolate the
sample. Nevertheless, if one had pure enol ether (via quickly
filtering through silica gel/chromatography), the reaction
went very cleanly using TMSOTf, which provided the best
yield and conversion to product; however, a full equivalent
was required as the resulting formation of methanol quenches
the catalyst. Second, if the ethyl ester of the indole was used,
transesterification of the esters gave complex reaction
mixtures. Reduction of the nitro group was accomplished
with SnCl2; however, the removal of the tin salt byproducts
gave unacceptable filtration rates (to be discussed in depth
in the Discussion). The E/Z ratio of the diester product was
approximately 80:20. Hydrolysis with LiOH gave crude
MDL 103371.
The first potential route utilizing the Stille Coupling4
(Scheme 1) was easily eliminated. The prerequisite inter-
mediates were coupled in the presence of bis(acetonitrile)
palladium dichloride in NMP at elevated temperatures. The
only analogue prepared via this route was the phenyl
derivative which was obtained in 30% yield after extensive
chromatographic purification. In addition to the product
purification issues, the use of the iodinated indole and vinyl
tin intermediate introduced waste issues and toxic handling
hazards.
The second potential route utilized a Suzuki Coupling5
as shown in Scheme 2. In this approach the indole aldehyde
(product from a Vilsmeier-Haack) was protected as the
N-tosylate. This was treated with an R-bromo-phosphonate
ester6 under Horner-Emmons conditions to give the desired
(Z)-vinyl bromide in 65% yield after fractional crystallization
from cyclohexane. Treatment of the vinyl bromide with a
wide variety of arylboronic acids under some modified
Suzuki conditions (method A,7a method B)7b-d gave the
desired mixed esters with variable results. It was hoped that
the coupled products would possess only the desired E-
stereochemistry, thus simplifying purification and charac-
terization of the products.
(2) (a) Salituro, F. G.; Baron, B. M.; Harrison, B. L.; Nyce, P. L. U.S. Patent
5,519,048, 21 May 1996. (b) Harrison, B. L.; Nyce, P. L.; Farr, R. A. U.S.
Patent 5,563,157, 8 October 1996. (c) Farr, R. A.; Harrison, B. L.; Nyce,
P. L. U.S. Patent 5,981,553, 9 November 1999.
(3) Baron, B. M.; Harrison, B. L.; Kehne, J. H.; Schmidt, C. J.; Gross, R. W.;
van Giersbergen, P.; White, H. S.; Farr, R.; Siegel, B. W.; Slone, A. L.;
Senyah, Y.; McCloskey, T. C.; Fadayel, G.; Taylor, V.; Murawsky, M.;
Meikrantz, S.; Nyce, P.; Janowick, D.; Salituro, F. G. Pharmacological
characterization of MDL 105,519, a systemically active antagonist of the
NMDA receptor-associated glycine recognition site. Eur. J. Pharmacol.
1997, 323, 181-192.
Purification of MDL 103371 could be accomplished by
either chromatography of the diester prior to hydrolysis or
by crystallization of 1 from 100 parts of methanol after
hydrolysis. The chromatography was tedious, and the
crystallization was problematic. It became clear from this
work that aside from any synthetic approach an improved
purification process was desperately needed. As a general
comment, the scope of this procedure may be limited to enol
ethers having electron-withdrawing or weak electron-donat-
ing groups, since the m- or p-methoxyphenyl- and 2- or
(4) (a) Stille, J. K.; Groh, B. L. J. Am. Chem. Soc. 1987, 109, 813-817. (b)
Stille, J. K. Angew. Chem., Intd. Ed. Engl. 1986, 25, 508-524.
(5) (a) Maddaford, S. P.; Keay, B. A. J. Org. Chem. 1994, 59, 6501-6503. (b)
Shieh, W.-C, Carlson, J. A. J. Org. Chem. 1992, 57, 379-381. (c) Miyaura,
N.; Ishiyama, T.; Sasaki, H.; Ishikawa, M.; Satoh, M.; Suzuki, A. J. Am.
Chem. Soc. 1989, 111, 314-312.
(6) McKenna, Charles E.; Khawli, Leslie A. J. Org. Chem. 1986, 51, 5467-
5471.
(7) (a) Hoshino, Y.; Miyaura, N.; Suzuki, A. Bull. Chem. Soc. Jpn. 1988, 61,
3008-3010. (b) Farina, V.; Krishnan, B. J. Am. Chem. Soc. 1991, 113,
9585-9595. (c) Thompson, W. J.; Jones, J. H.; Lyle, P. A.; Thies, J. E. J.
Org. Chem. 1988, 53, 2052-2055. (d) Wallow, T. I.; Novak, B. M. J. Org.
Chem. 1994, 59, 5034-5037.
478
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development

Scheme 2
Scheme 3
3-thienyl-substituted enol ethers did not appear to give any
coupled products. Some of the analogues prepared were the
p-F, p-Cl, p-Br, p-I, p-Me, p-CF3, m-NO2, and o-Cl deriva-
tives.
Scheme 4
Since the enol ether method was not applicable to the
synthesis of analogues with electron-donating phenyl groups,
the Knoevenagel condensation8 was investigated (fourth
approach shown in Scheme 4). Under classical Knoevenagel
condensation conditions arylacetic esters are condensed with
aldehydes and ketones in the presence of a catalytic amount
of base to give the R,â-unsaturated esters. Any attempts with
multiple analogues of arylacetic esters were not successful;
however, the condensation of the appropriate arylacetonitrile
with the aldehyde gave the desired results. Depending on
the aryl substituents, the condensation was rather sluggish
and proceeded in low yield with starting material being
recovered.
The drawback of this method, with exception of the
pyridyl analogues, was that the hydrolysis of the R,â-
unsaturated nitriles proceeded only to the amide stage.
Conditions for further or direct hydrolysis to the diacids had
not been fully investigated during the course of this work.
The commercially available m-nitrophenylacetonitrile made
(8) (a) Contracted to Wynberg H. Syncom, Holland. (b) Crowell, T. I.; Kim,
T.-R. J. Am. Chem. Soc. 1973, 95:20, 6781-6786. (c) Prout, F. S.;
Beaucaire, V. D.; Drykacz, G. R.; Koppes, W. M.; Kuznicki, R. E.;
Marlewski, T. A.; Pienkowski, J. J.; Puda, J. M. J. Org. Chem. 1973, 38,
1512-1517. (d) Crowell, T. I., Peck, D. W. J. Am. Chem. Soc. 1953, 75,
1075. (e) Cordes, E. H., Jencks, W. P. J. Am. Chem. Soc. 1962, 84, 826. (f)
For more Knoevenagel references, see: House, H. O. Modern Synthetic
Reactions, 2nd ed.; The Benjamin/Cummings Publishing Company: New
York, 1972.
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
479

Scheme 5
Scheme 6
this attractive approach to 1 a high priority. Some of the
other analogues prepared were the p-NO2, 2-, 3-, and
4-pyridyl derivatives.
Chemical Development’s initial efforts were focused on
the enol ether route and the Knoevenagel condensation.
Although the enol ether route was not attractive from a scale-
up position, it provided us with problems common to the
two routes (nitro reduction and purification). The Knoe-
venagel route was the one that ultimately provided the
solution to preparing larger quantities of MDL 103371. From
this point on in the discussion the focus will be directed
towards development of the Knoevenagel approach.
To circumvent the problems encountered with the hy-
drolysis of the nitriles, a fifth approach utilizing the Mu-
kaiyama reaction9 was investigated (shown in Scheme 5). It
was felt that the use of the Mukaiyama reaction, which
involves transfer of a silyl group to the oxygen of the aldol,
could circumvent the reversibility of the Knoevenagel
condensation with the esters and drive the reaction to
completion. In addition, the Mukaiyama reaction would also
provide a means of preparing analogues from both alkyl-
and arylacetic esters containing relatively unactivated me-
thylene groups.
In practice,10 the reaction gave complex mixtures of
erythro/threo aldols and their silyl-protected counterparts
which were difficult to separate via chromatography. Second,
it was difficult to control the retro aldol reaction from taking
place during the elimination process, thus giving unrepro-
ducible results. This approach received low priority. Some
of the analogues prepared were the phenyl, m-pyridyl,
m-phenoxyphenyl, and m-methoxyphenyl derivatives.
Discussion
Preparation of the indole aldehyde precursor to the
Knoevenagel reaction utilized classical chemistry (Scheme
6). 3,5-Dichlorophenylhydrazine (2) was treated with ethyl
pyruvate in ethanol to give (E/Z)-ethyl-2-[(3,5-dichlorophen-
yl)hydrazono]propanoate (3) in 89% yield. The reaction was
originally performed with the addition of sulfuric acid as
the cataylst; however, it was later found that the hydrochloric
acid from the 3,5-dichlorophenylhydrazine hydrochloride was
a sufficient acid source. Hydrazone 3 was cyclized under
Fischer Indole cyclization11 conditions with polyphosphoric
acid (PPA) to give ethyl-4,6-dichloro-1H-indole-2-carboxyl-
ate (4) in 97% yield. The viscous nature of PPA at ambient
temperature made it difficult to stir with a glass stirring rod
in a 22 L flask, and therefore toluene was added to reduce
(9) Mukaiyama, T.; Matsui, S.; Kashiwagi, K. Chem. Lett. 1989, 993-996.
(10) (a) Prepared by the method of Emde, H.; Simchen, G. Synthesis 1977, 867-
869. (b) Aaberg, A.; Gramstad, T.; Husebye, S. Tetrahedron Lett. 1979,
2263-2264. (c) Hendrickson, J. B.; Schwartzman, S. M. Tetrahedron Lett.
1975, 277-280
(11) Loudon, M. G. Organic Chemistry; Addison-Wesley Publishing Com-
pany: Reading, MA, 1983.
480
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development

Scheme 7
Figure 1. ReactIR 3D plot of Knoevenagel reaction.
the viscosity. The toluene did not effect the reaction and
could be removed prior to the quench via distillation. In a
reactor, however, stirring was found adequate without the
need for toluene. The product of the Fischer indole cycliza-
tion 4 was subjected to a Vilsmeier-Haack11 formylation.
Originally dichloroethane (DCE) was used with the phos-
phorus oxychloride to form and isolate the Vilsmeier-Haack
intermediate as the dichlorophosphoric acid salt. It was found
toluene was an efficient and acceptable replacement for the
environmentally undesirable DCE. The Vilsmeier-Haack
intermediate was then treated with buffered water to give
ethyl-4,6-dichloro-3-formyl-1H-indole-2-carboxylate (5) in
88% yield. It was later learned that the toluene could be
removed via distillation, and the hydrolysis could be
performed on the mixture without having to isolate the
intermediate via filtration. The entire Fischer indole/Vils-
meier-Haack process was tried in toluene. However, there
were impurities formed during the hydrazone formation that
could not be removed in intermediates 3, 4, or 5.
Ethyl-4,6-dichloro-3-formyl-1H-indole-2-carboxylate (5)
was treated with 3-nitrophenylacetonitrile (Scheme 7) under
Knoevenagel8 conditions using catalytic piperidine in ethanol
to give ethyl-(Z)-4,6-dichloro-3-[2-cyano-2-(3-nitrophenyl)-
ethenyl]-1H-indole-2-carboxylate (6) exclusively as the Z
isomer in 91% yield (large NOEs from H-9 to H-12 and
H-16 are consistent with (Z)-olefin geometry). Product began
to crystallize from the hot mixture approximately 15 min
after reaching refluxing temperature. After the mixture
cooled, the product 6 was isolated with greater than 99%
purity. The insoluble nature of 6 most likely helps drive the
reaction to completion and eliminates any “retro aldol”
condensation from taking place. However, because of the
insolubility of nitrile 6, we initially could not find an efficient
way to monitor the progress of the reaction.
The availability of new instrumentation allowed us to
investigate the Knoevenagel reaction using in situ FTIR
monitoring techniques (ReactIR)12 The data generated is
summarized in three figures , including reaction profile 3D
plot (Figure 1), ConcIRT analysis of identified components
(Figure 2), and reaction profile of identified components
(Figure 3).
(12) In situ FTIR reaction monitoring was performed using a ReactIR MP Mobile
Reaction Analysis System (ASI Applied Systems, Millersville, MD)
equipped with a 9-bounce, SiComp probe. The probe was inserted into glass
round-bottomed flasks via a 24/40 airtight adapter and submersed into the
reaction mixtures. (a) First, the progress of the reaction was monitored by
FTIR in real time by collecting data (4000-650 cm-1) at predetermined
intervals (typically every 0.5-2 min) for the duration of the reaction. The
set of absorbance spectra was then plotted as a function of time to generate
a ReactIR 3D Plot profile for each reaction sequence. (b) Second, each
ReactIR 3D Plot that was generated was then analyzed using ConcIRT, a
post-experiment algorithm which tracks changes in absorbance profiles that
occur over time as a reaction proceeds. Even with complex reaction mixtures
that have overlapping IR bands, ConcIRT has the capability of “extracting”
absorption bands of the individual species in a reaction mixture, and thus
can provide the individual IR spectrum for each “identified” component,
along with the corresponding reaction concentration profiles.
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
481

Figure 2. ConcIRT analysis of the Knoevenagel reaction.
Figure 3. ConcIRT component profiles of Knoevenagel reaction.
The ConcIRT algorithm looked at the changing bands in
the IR plot and identified three components in the reaction
mixture (Figure 2). Upon closer analysis, those components
were the two starting materials and product (as compared to
blank runs of each component). Upon inspection of the
components’ concentration versus time (Figure 3) plot it
became clear that the reaction was over in about 1.5 h. The
first 15 min were used to heat the reaction to reflux. During
that time, the 3-nitrophenylacetonitrile was not completely
in solution as evident from Figure 3 (solution concentration
increases during heat up, then decreases as it is consumed).
The concentrations of ethyl-4,6-dichloro-3-formyl-1H-indole-
2-carboxylate (5) and 3-nitrophenylacetonitrile began to
decrease, and ethyl-(Z)-4,6-dichloro-3-[2-cyano-2-(3-nitro-
phenyl)ethenyl]-1H-indole-2-carboxylate (6) began to appear
and increase in the first 30 min. However, due to the
insoluble nature of 6, the apparent concentration decreases
as it drops out of solution due to the inability of the ReactIR
probe to take an IR of a suspended solid. Nevertheless, one
can easily monitor the disappearance of the two starting
materials to determine the progress of the reaction. It may
be interesting to note that it is not evident if the Z isomer is
formed exclusively or if the E isomer is equilibrated to the
thermodynamically favored Z isomer.
Although the reduction of ethyl-(Z)-4,6-dichloro-3-[2-(3-
aminophenyl)2-cyanoethenyl]-1H-indole-2-carboxylate (7)
with SnCl2 went in high chemical purity, the yield of the
reaction varied (67-97%) due to emulsion problems during
the work-up. Attempted filtration of the emulsion through a
pad of Celite prior to separation of the aqueous and organic
layers was tedious due to poor flow rate and difficulty in
extracting all of aniline 7 from the filter cake. Because of
these problems, as well as the toxicity issues surrounding
the use of tin reagents, alternative reaction conditions were
sought for the reduction of nitro-nitrile 6.
There is a veritable plethora of methods for reduction of
aromatic nitro compounds to the corresponding anilines;
however, nitro-nitrile 6 demands a very selective reducing
agent because of the breadth of functionality present in the
molecule (i.e., the nitrile, the aryl chlorides, and the ester
moieties). A widely used reagent for the reduction of nitro
482
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development

Scheme 8
compounds to amines is iron.13 Most of these reactions use
an iron salt in conjunction with iron metal to accomplish
the reduction. Thus, nitro-nitrile 6 was treated with Fe/
products. Because it was thought palladium catalysts were
possibly causing the indole ring of nitro-nitrile 6 to undergo
hydro-dehalogenation, PtO2 and Pt/C were examined. Reduc-
tion of nitro-nitrile 6 with PtO2 under one atmosphere of
hydrogen gave two products (TLC), one of which appeared
to be desired product aniline 7. A more interesting result
was obtained using Pt/C with hydrazine to reduce nitro-nitrile
6. A single product was obtained which was initially thought
to be aniline 7; however, careful comparison of this new
product with aniline 7 by TLC and NMR revealed they were
not the same. The new product was identified as the
hydroxylamine analogue. It is interesting to note that this
new product is not formed when hydrogen (1 atm) is
substituted for hydrazine, or if hydrazine is present but the
catalyst is omitted. The hydroxylamine was stable at ambient
temperature under inert atmosphere as a solid, but when it
was dissolved in either THF or EtOAc and allowed to stand
at ambient temperature, it decomposed into primarily one
new unidentified compound.
The best conditions found to date for the reduction of
nitro-nitrile 6 to aniline 7 use Rh/Al2O3 and Et3N under an
atmosphere of hydrogen in THF at ambient temperature for
∼20 h. Under these conditions a near quantitative yield of
aniline 7 is obtained. It was not ascertained if Et3N was
absolutely crucial for the reaction.19 The reaction was
monitored by HPLC, and an intermediate product could be
seen, which was converted to product if the reaction was
allowed to run to completion. This intermediate product is
thought to be hydroxylamine since it had the same HPLC
retention time as a previously identified hydroxylamine
sample. Varying the pressure of hydrogen from atmospheric
to 40 psi resulted in complete reaction in <16 h with no
loss of product purity. However, upon scaling the reaction
14
FeSO4 in various solvent systems (EtOH/H2O, THF/H2O,
and EtOAc/H2O) heated at reflux. All of the reactions gave
similar resultsslittle to no reaction with poor chemoselec-
tivity for formation of aniline 7. The use of Fe/NH4Cl15 in
MeOH/H2O heated at reflux also showed little reaction. An
attempt was made to reduce nitro-nitrile 6 with Cu(acac)2/
NaBH4 in EtOH/THF,16 but the reaction did not go to
completion and gave several products. A general problem
with these attempted reductions was the poor solubility of
nitro-nitrile 6 in most organic solvents. It was found that 6
was soluble in DMSO and THF ( ∼1 g of nitro-nitrile 6 per
10-15 mL of THF at ambient temperature) and slightly
soluble in EtOAc but insoluble in chlorinated solvents,
MeCN, alcohols,17 heptane, or toluene.
The next approach to reduce aromatic nitro-nitrile 6 was
via catalytic hydrogenation (Scheme 8). The choice of
catalyst was deemed to be crucial to the success or failure
of this approach because of the various functionality present
in nitro-nitrile 6. The use of Pd/C with hydrazine had been
reported to reduce aromatic nitro groups in the presence of
aliphatic nitriles;18 however, under these conditions, nitro-
nitrile 6 afforded aniline 7 contaminated with two unidenti-
fied by-products. Use of Pd/Pb/C with hydrazine gave several
(13) Hodgson, H. H.; Whitehurst, J. S. J. Chem. Soc. 1945, 202.
(14) Hodgson, H. H,; Hathway, D. E. J. Chem. Soc. 1944, 538.
(15) Ramadas, K.; Srinivasan, N. Synth. Commun. 1992, 22, 3189; b) Tsuji, K.;
Nakamura, K.; Konishi, N.; Okumura, H.; Matsuo, M. Chem. Pharm. Bull.
1992, 40, 2399.
(16) Kizil, M.; Lampard, C.; Murphy, J. A. Tetrahedron Lett. 1996, 37, 2511.
For other NaBH4-metal salts systems see: He, Y.; Zhao, H.; Pan, X.; Wang,
S. Synth. Commun. 1989, 19, 3047 and references therein.
(17) The solubility of 6 in methanol was not checked due to concerns of
transesterification.
(18) Adger, B. M.; Young, R. G. Tetrahedron Lett. 1984, 25, 5219. The use of
Raney nickel with hydrazine affords concomitant reduction of the nitrile to
the corresponding amine but leaves aromatic bromides and chlorides
unchanged (also see: Leggetter, B. E.; Brown, R. K. Can. J. Chem. 1960,
38, 2363).
(19) Amine was added to the reaction to retard possible hydrodechlorination of
the indole. An initial experiment done without Et3N had a significant by-
product produced, which is now thought to be hydroxylamine based upon
HPLC retention times, so it may be possible to perform the reaction without
Et3N.
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
483

Scheme 9
with an increase in pressure and temperature, one began to
see a des-chloro analogue.20 This impurity is difficult to
remove in any of the remaining steps, thus contaminating
the final product.
was confirmed by subjecting a clean sample of 1 to prolonged
basic hydrolysis conditions and monitoring the formation of
the des-chloro analogue as compared to an authentic sample.
The mechanism of this unfortunate phenomenon was not
investigated nor is it fully understood.
The hydrolysis to the acid was accomplished via the amide
in a two-step sequence. First, treatment of ethyl-(Z)-4,6-
dichloro-3-[2-(3-aminophenyl)2-cyanoethenyl]-1H-indole-2-
carboxylate (7) with a 1:1 mixture of sulfuric and acetic acid
gave the amide (one time purification and isolation yield of
95%). Many acid-acid, acid-water combinations were
screened only to produce decomposition and multiple product
mixtures. If the sulfuric acid-acetic acid hydrolysis tem-
perature was kept at 50 °C, the Z geometry of the olefin
was retained. However, if the temperature of the hydrolysis
were increased to 90 °C, then mixtures of E/Z isomers were
obtained in variable ratios depending on the time at the
elevated temperatures. Nevertheless, the product amide-
aniline 8 was subjected to potassium hydroxide to give the
desired crude MDL 103371 in 83% yield. The ratio of the
crude product 1 was approximately 80:20 E/Z and could not
be equilibrated any further to the desired E isomer under
many conditions. Second, subjecting different mixtures of
the E/Z to the basic hydrolysis conditions consistently drove
the mixture to approximately 80:20 ratio of E/Z isomers, the
apparent thermodynamic equilibrium ratio under these given
conditions. Much to our surprise, the basic hydrolysis
produced small amounts of the des-chloro impurity.20 This
Purification
Two purification techniques were ultimately identified.
The first was found by allowing the potassium hydroxide
hydrolysis reaction mixture to slowly cool overnight. Crystals
of the dipotassium salt 9 slowly precipitated from the reaction
medium. This solid could be isolated by filtration and was
determined to have roughly 93% of the E isomer and 4% of
the Z isomer. The salt could be neutralized with hydrochloric
acid to give back the diacid under mild conditions which
did not to affect this ratio.
The second purification was the result of parallel crystal-
lization studies. Crystallization of a 85:15 E/Z mixture from
acetone gave crystals of the acid-catalyzed Schiff base 10
(Scheme 9) in greater than a 99.9:0.1% E/Z ratio. (This was
a sample that had not been purified as the dipotassium salt,
9; however, 9 had been over-acidified from the work-up of
potassium hydrolysis, and thus had HCl present. Typical E/Z
ratios before the dipotassium salt purification were on the
order of 85:15 E/Z.) The Schiff base was easily converted
(20) The des-chloro isomer was isolated by preparative chromatography and
characterized to be the following structure. The other possible des-chloro
isomers were not detected with LC/MS techniques.
484
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development

back to MDL 103371 under mild conditions which did not
affect the desired ratio. The product was isolated in the
absence of light in a Nutsche filter and stored appropriately
without further isomerization. Although the des-chloro
analogue could not be fully removed by either of the two
purification methods, it could be minimized by repeated
Schiff base crystallization.
dichlorophenylhydrazine hydrochloride (Acros) and 12 L of
EtOH was charged to a 22-L three-necked flask fitted with
a stirrer, thermometer, condenser, and a continuous N2 purge.
The mixture was stirred at 50 °C for 1 h (there were
insolubles present). A total of 1.08 kg (9.37 mol) of ethyl
pyruvate was added in one portion. The temperature rose
from 45 to 54 °C over ∼5 min. The reaction mixture was
allowed to stir at ambient temperature. The reaction was
monitored by GC (tR of 2 ) 9.2 min; tR of 3 ) 11.3 min)
and reached completion after 4 h. The insolubles were filtered
off and washed with 1 L of EtOH. The solvent was removed
from the filtrate at 40 °C/50 Torr to give a residue. The solid
obtained was air-dried to a constant weight to give 2.282
kg, 89% yield, of (E/Z)- ethyl 2-[(3,5-dichlorophenyl)-
Conclusions
The preparation of kilogram quantities of MDL 103371
utilized some classical chemistry. Fischer indole cyclization,
followed by a Vilsmeier-Haack formylation and a Knoe-
venagel condensation gave immediate access into the proper
carbon framework of the target molecule. A unique hydro-
genation catalyst and solvent system followed by a two-step
acid-base hydrolysis gave the crude product. Purification
was accomplished by potassium salt crystallization followed
by Schiff base formation to give MDL 103371 in nine steps
(including the purification step) in an overall yield of 38%.
All of the steps gave yields of greater than 80% (except the
Schiff base, 78%) with easily isolated solids. (Two batches
of 1.3 and 2.0 kg of MDL 103371 were prepared in
Cincinnati.)
1
hydrazono]propanoate, 3. H NMR (DMSO-d6) δ 10.1 (bs,
1 H), 7.21 (s, 2 H), 6.98 (s, 1 H), 4.21 (q, 2 H, J ) 7.2 Hz),
2.05 (s, 3 H), 1.26 (t, 3 H, J ) 7.1 Hz). 13NMR (DMSO-d6)
164.7, 147.0, 135.1, 134.7, 119.6, 112.1, 60.6, 14.1, 12.1.
Anal. Calcd for C11H12Cl2N2O2: C, 48.02; H, 4.40; N, 10.18.
Found: C, 47.43; H, 4.25; N, 9.95. MS m/z: (M+) calcd
274.0, obsd (M - H) 273.8. IR (2% in KBr) Vmax/cm-1
)
1698, 1583.
Ethyl 4,6-Dichloro-1H-indole-2-carboxylate, 4. A mix-
ture of 53 kg of polyphosphoric acid, 14 L of toluene, and
2.83 kg (10.3 mol) of ethyl 2-[(3,5-dichlorophenyl)hydra-
zono]propanoate, 3 was charged to a 20 gallon glass-lined
reactor under a continuous N2 purge. The mixture was heated
to 95-100 °C (NOTE: At ∼70 °C HCl gas was given off).
The progress of the reaction was monitored by GC (tR of 4
) 11.8 min, tR of 3 ) 11.3 min) and found to be complete
after 2 h. The reaction mixture was cooled to 80 °C, and
toluene was removed at 100 Torr. The reaction mixture was
discharged from the reactor and quenched into a 30 gallon
Hastelloy reactor which contained 50 kg of ice and 10 L of
ice water. The resulting slurry was stirred at 35-50 °C for
30 min. The product was filtered off, washed acid free with
2 × 20 L H2O, then air-dried to give 2.57 kg, 97% yield, of
ethyl 4,6-dichloro-1H-indole-2-carboxylate, 4. 1H NMR
(DMSO-d6) δ 12.46 (bs, 1 H), 7.45 (s, 1 H), 7.28 (s, 1 H),
7.11 (s, 1 H), 4.41 (q, 2 H, J ) 7.1 Hz), 1.34 (t, 3 H, J )
7.1 Hz). 13NMR (DMSO-d6) 160.8, 137.8, 129.4, 129.3,
127.1, 124.5, 120.2, 111.5, 105.5, 61.0, 14.2. Anal. Calcd
for C11H9Cl2NO2: C, 51.19; H, 3.51; N, 5.43. Found: C,
51.34; H, 3.38; N, 5.34. MS m/z: (M+) calcd 257.0, obsd
(M - H) 256.5 IR (2% in KBr) Vmax/cm-1 ) 1698.
Ethyl 4,6-Dichloro-3-formyl-1H-indole-2-carboxylate,
5. A mixture of 2.28 kg (8.85 mol) of ethyl 4,6-dichloro-
1H-indole-2-carboxylate, 4, 5.8 L of 1,2-dichloroethane, and
972 g (13.3 mol) of DMF was charged to a 22-L three-necked
flask fitted with a stirrer, thermometer, condenser, and
continuous N2 purge. [NOTE: Toluene was used to replace
dichloroethane (DCE) in future experiments (same volume).
Because the full analytical data was collected on the DCE
prepared sample, the experimental will reflect that exact
experiment. The toluene process was developed by our Dow
Chemical colleagues.] The mixture was stirred to a smooth
slurry at 24 °C and 2.04 kg (13.3 mol) of POCl3 was added
over 2 min. The temperature rose from 24 to 58 °C. The
For the process to be complete, a better understanding of
the hydrogenation and the basic hydrolysis would be neces-
sary to control des-chloro formation. Because the project was
terminated, these issues were not investigated and are still
not fully understood.
Experimental Section
1
General. All H- and 13C NMR spectra were performed
on a Varian XL-300 spectrophotometer at 300 and 75 MHz,
respectively. All GC data were collected on a Hewlett-
Packard 5890 GC with an HP 5972 mass selective detector.
The GC column was an HP-5 (cross-linked 5% Ph Me
silicone, 30 m × 0.25 mm × 0.25 µm film thickness);
injector temperature - 250 °C; detector temperature - 280
°C; temperature program - 50 °C for 1 min then heated to
300 °C at the rate of 20 °C/min. HPLC analysis was done
on a Thermo Separations P2000 HPLC with a Spectra 100
detector and a Spectra Physics SP4270 integrator The column
used was a Waters Symmetry C18 column (3.9 × 150 mm);
the wavelength of detection was at 230 nm; the mobile phase
used for the chromatography to monitor the synthesis of (E)-
4,6-dichloro-3-[2-(3-aminophenyl)-2-carboxyethenyl]-1H-in-
dole-2-carboxylic acid and (E)-4,6-dichloro-3-[2-carboxy-
2-[3-[(1-methylethylidene)amino]ethenyl]-1H-indole-2-car-
boxylic acid monohydrochloride was 70% 0.05 M NaH2-
PO4 buffer pH 3.0/30% CH3CN at a flow rate of 1.0 mL/
min. The mobile phase used to monitor the synthesis of ethyl
(E/Z)-4,6-dichloro-3-[3-amino-2-aminophenyl)-3-oxo-1-pro-
penyl]-1H-indole-2-carboxylate and ethyl (Z)-4,6-dichloro-3-[2-
(3-aminophenyl)2-cyanoethenyl]-1H-indole-2-carboxylate was
40% 0.05% aqueous TFA/60% CH3CN at a flow rate of 1.0
mL/min. Thin-layer chromatography was done using 5 ×
10 cm Merck 60F-254 plates.
(E/Z)- Ethyl 2-[(3,5-Dichlorophenyl)hydrazono]pro-
panoate, 3. A mixture of 2.0 kg (9.36 mol) of 3,5-
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
485

solution was heated to and held at reflux (88 °C) for 18 h
[90 °C for toluene]. The progress of the reaction was
monitored by GC (tR of 5 ) 13.3 min, tR of 4 ) 11.8 min)
and typically ∼8% of MDL 102317 (4) remained. The slurry
was cooled to and held at 25 °C for 30 min. The precipitate
was filtered off and washed with 2 × 2 L of 1,2-dichlo-
roethane and 2 L of Et2O [Et2O was replaced with ethanol
in the toluene process]. The filter cake was slurried in 39 L
of 1 M NaOAc for 3 h. The product was filtered off and
washed with 2 × 20 L of H2O, 5 L of Et2O, then air-dried
for 3 d to give 2.21 kg, 88% yield, of ethyl 4,6-dichloro-3-
formyl-1H-indole-2-carboxylate, 5. 1H NMR (DMSO-d6) δ
13.00 (bs, 1 H), 10.60 (s, 1 H), 7.53 (s, 1 H), 7.37 (s, 1 H),
4.45 (q, 2 H, J ) 7.1 Hz), 1.39 (t, 3 H, J ) 7.1 Hz). (DMSO-
d6) 160.5, 148.7, 139.2, 137.5, 134.6, 131.9, 131.3, 130.0,
127.2, 124.2, 122.3, 122.2, 120.3, 116.9, 115.8, 114.4, 112.0,
61.5, 14.0.Anal. Calcd for C12H9Cl2NO3: C, 50.38; H, 3.17;
N, 4.90. Found: C, 50.23; H, 3.06; N, 4.85. MS m/z: (M+)
calcd 285, obsd (M + H) 286.4. IR (2% in KBr) Vmax/cm-1
) 1730, 1667.
Ethyl (Z)-4,6-Dichloro-3-[2-cyano-2-(3-nitrophenyl)-
ethenyl]-1H-indole-2-carboxylate, 6. A mixture of 59 L of
EtOH, 2.21 kg (7.76 mol) of ethyl 4,6-dichloro-3-formyl-
1H-indole-2-carboxylate, 1.26 kg (7.76 mol) of 3-nitrophen-
ylacetonitrile, and 205 g (2.4 mol) of piperidine was charged
to a 30 gallon Hastelloy reactor under a continuous N2 purge.
The resulting mixture was heated to and held at reflux (79
°C) for 70 h. The reaction mixture was cooled to 25 °C.
The precipitate was filtered off and washed with 2 × 5 L of
EtOH. The product was air-dried to give 3.03 kg, 91% yield,
of ethyl (Z)-4,6-dichloro-3-[2-cyano-2-(3-nitrophenyl)ethen-
yl]-1H-indole-2-carboxylate, 6. 1H NMR (DMSO-d6) δ 12.95
(bs, 1 H), 8.63 (s, 1 H), 8.49 (dd, 1 H, J ) 1.9, 1.9), 8.32
(dd, 1 H, J ) 8.1, 1.9), 8.23 (dd, 1 H, J ) 8.0, 1.9), 7.85
(dd, 1 H, J ) 8.1, 8.0), 7.54 (d, 1 H, J ) 1.8), 7.36, (d, 1 H,
J ) 1.8), 4.35 (q, 2 H, J ) 7.1 Hz), 1.26 (t, 3 H, J ) 7.1
Hz). 13NMR (DMSO-d6) 160.4, 148.4, 138.9, 137.2, 134.5,
131.7, 131.2, 129.7. 127.9, 126.4, 123.7, 122.0 (2), 120.1,
115.9, 155.7, 114.2, 111.9, 61.42, 13.9. Large NOEs from
the H-9 to both H-12 and H-16 are consistent with the Z
geometry. Anal. Calcd for C20H13Cl2N3O4: C, 55.83; H, 3.05;
N, 9.77. Found: C, 55.69; H, 3.07; N, 9.66. MS m/z: (M+)
calcd 429, obsd (M - H) 428.6. IR (2% in KBr) Vmax/cm-1
) 1708.
washed with 2 × 1 L of THF. The filtrate was concentrated
at 35 °C/50 Torr and the solid obtained was dried at 30 °C/
50 Torr for 18 h to give 813.0 g, 99% yield, of ethyl (Z)-
4,6-dichloro-3-[2-(3-aminophenyl)2-cyanoethenyl]-1H-indole-
1
2-carboxylate, 7. H NMR (DMSO-d6) δ 12.95 (bs, 1 H),
8.09 (s, 1 H), 7.51 (d, 1 H, J ) 1.7), 7.31 (d, 1 H, J ) 1.7),
7.15 (dd, 1 H, J ) 8.0,8.1), 6.93 (dd, 1 H, J ) 1.7,1.7),
6.87(dd, 1 H, J ) 8.2, 1.7), 6.65, (dd, 1 H, J ) 8.2, 1.7),
5.39 (s, 2 H), 4.32 (q, 2 H, J ) 7.1 Hz), 1.26 (t, 3 H, J )
7.1 Hz). 13NMR (DMSO-d6)160.7, 149.7, 137.4, 134.5,
133.8, 129.9, 129.8, 127.3, 126.9, 122.3, 122.0, 118.6, 117.7,
115.1, 115.0, 113.0, 111.8, 110.8, 61.3, 14.0. Anal. Calcd
for C20H15Cl2N3O2: C, 60.02; H, 3.78; N, 10.50. Found: C,
60.10; H, 3.81; N, 10.29. MS m/z: (M+) calcd 399.0, obsd
(M + H) 400.4. IR (2% in KBr) Vmax/cm-1 ) 1708.
Ethyl (E/Z)-4,6-Dichloro-3-[3-amino-2-aminophenyl)-
3-oxo-1-propenyl]-1H-indole-2-carboxylate, 8: A mixture
of 1.52 kg (3.79 mol) of ethyl (Z)-4,6-dichloro-3-[2-(3-
aminophenyl)2-cyanoethenyl]-1H-indole-2-carboxylate, 7,
and 4.85 L of acetic acid was charged to a 22-L three-necked
flask fitted with a stirrer, thermometer, dropping funnel, and
continuous N2 purge. The mixture was stirred to a smooth
slurry and a total of 4.85 L of H2SO4 was added over 3-5
min. The temperature rose from 24 to 90 °C. The reaction
temperature was held at 90 °C. The progress of the reaction
was monitored by HPLC (tR of ethyl (Z)-4,6-dichloro-3-[2-
(3-aminophenyl)2-cyanoethenyl]-1H-indole-2-carboxylate, 7s
5.8 min, tR of ethyl (E/Z)-4,6-dichloro-3-[3-amino-2-
aminophenyl)-3-oxo-1-propenyl]-1H-indole-2-carboxylate, 8s
1.2 min) and found to be complete after 1 h. The reaction
mixture was cooled to 50 °C and quenched by pouring the
reaction mixture into 22 kg of ice. The pH was adjusted to
4.9 by addition of 33 L of 6 M NaOH while maintaining a
reaction temperature of 40-50 °C. The precipitate which
formed was filtered off and washed with 3 × 10 L of H2O.
The product was air-dried to remove a majority of the H2O,
but the MDL 46757 obtained was used as a water wet pulp
in the subsequent step. The yield for the reaction was 1.68
kg, 106% yield, of crude ethyl (E/Z)-4,6-dichloro-3-[3-
amino-2-aminophenyl)-3-oxo-1-propenyl]-1H-indole-2-car-
boxylate, 8. This was used without further purification.
(E)-4,6-Dichloro-3-[2-(3-aminophenyl)-2-carboxyethen-
yl]-1H-indole-2-carboxylic acid, 1. A mixture of 44.5 L of
6 M KOH and 1.34 kg (3.2 mol) of ethyl (E/Z)-4,6-dichloro-
3-[3-amino-2-aminophenyl)-3-oxo-1-propenyl]-1H-indole-2-
carboxylate (NOTE: 1.55 kg of water wet pulp was used)
was charged to a 20 gallon glass-lined reactor. The mixture
was heated to and held at reflux (108 °C) for 3.5 h (∼1h
after complete solution). The progress of the reaction was
carefully monitored by HPLC (tR of (E)-4,6-dichloro-3-[2-
(3-aminophenyl)-2-carboxyethenyl]-1H-indole-2-carboxyl-
ic acid 1 ) 7.5 min, tR of Z isomer of MDL 103371, 1 )
9.9 min, tR of 8 ) 13.6 min, tR of Z isomer of des-chloro
impurity ) 3.0 min) and found to be 76% (E)-4,6-dichloro-
3-[2-(3-aminophenyl)-2-carboxyethenyl]-1H-indole-2-car-
boxylic acid, and 20% Z isomer. The reaction mixture was
cooled to 25 °C over 2 h and held at this temperature for 20
min. The precipitate was filtered off and not washed. The
Ethyl (Z)-4,6-Dichloro-3-[2-(3-aminophenyl)2-cyanoeth-
enyl]-1H-indole-2-carboxylate, 7. A mixture of 871 g (2.03
mol) of ethyl (Z)-4,6-dichloro-3-[2-cyano-2-(3-nitrophenyl)-
ethenyl]-1H-indole-2-carboxylate, 13.5 L of THF, 175 g of
5% Rh/Al2O3, and 244 mL of triethylamine was charged to
a 5 gallon autoclave under a N2 blanket. The H2 pressure
was increased to about 50 psi and maintained at this pressure
throughout the reaction. The reaction was slightly exothermic
and the temperature was maintained at 23-36 °C by
occasional cooling. The progress of the reaction was
monitored by HPLC (tR of 7 ) 2.9 min, tR of 6 ) 14.8 min,
tR of hydroxylamine intermediate ) 4.4 min) and found to
be complete after 6 h (NOTE: H2 uptake ceased after ∼5
h). The catalyst was filtered off on a bed of dicalite and
486
•
Vol. 4, No. 6, 2000 / Organic Process Research & Development

di-potassium salt, 9 was analyzed by HPLC and found to be
93% (E)-4,6-dichloro-3-[2-(3-aminophenyl)-2-carboxyethen-
yl]-1H-indole-2-carboxylic acid, 4% Z isomer, and 0.9% des-
chloro. The dipotassium salt, 9 of (E)-4,6-dichloro-3-[2-(3-
aminophenyl)-2-carboxyethenyl]-1H-indole-2-carboxylic acid
was dissolved in 35 L of H2O, and the pH was 12.8. The
pH of the solution was adjusted to 3.2 by addition of 3.1 L
of 3 M HCl while maintaining a reaction temperature of
about 25 °C. The resulting slurry was stirred for 10 min.
The product was filtered off and washed with 2 × 10 L of
H2O. The material was dried at 50 °C/50 Torr to give 1.05
kg, 83% yield, of crude (E)-4,6-dichloro-3-[2-(3-aminophen-
yl)-2-carboxyethenyl]-1H-indole-2-carboxylic acid, 1.
Purification via (E)-4,6-Dichloro-3-[2-carboxy-2-[3-[(1-
methylethylidene)amino]ethenyl]-1H-indole-2-carboxyl-
ic Acid Monohydrochloride (10). A mixture of 1017 g (2.6
mol) of crude (E)-4,6-dichloro-3-[2-(3-aminophenyl)-2-car-
boxyethenyl]-1H-indole-2-carboxylic acid and 14 L of
acetone was charged to a 22-L three-necked flask fitted with
a stirrer, thermometer, condenser, and continuous N2 purge.
The mixture was heated to and held at reflux (56 °C) for 30
min. The resulting turbid solution was clarified through a
dicalite bed directly into a 22-L three-necked flask, and the
filter cake was washed with 2 L of acetone. The filtrate was
heated to and held at 50 °C. A total of 200 mL of
concentrated HCl was added over 5 min. The reaction
mixture was allowed to cool to rt over 2 h. The precipitate
which formed was filtered off and washed with 2 × 1 L of
acetone. The solid was dried at 25 °C/50 Torr to give 989.2
g, 78% yield of (E)-4,6-dichloro-3-[2-carboxy-2-[3-[(1-
methylethylidene)amino]ethenyl]-1H-indole-2-carboxylic acid
monohydrochloride, 10. The product contained <0.1% of Z
isomer by HPLC analysis.
for 15 min. The product was filtered off, washed with 4 ×
5 L of H2O, then dried for 24 h at 60 °C/50 Torr to give
1.97 kg, 89% yield, of (E)-4,6-dichloro-3-[2-(3-aminophen-
yl)-2-carboxyethenyl]-1H-indole-2-carboxylic acid, 1. 1H
NMR (DMSO-d6) δ 12.10 (s, 1 H), 7.98 (s, 1 H), 7.35 (s, 1
H), 7.18 (s, 1H,), 6.62 (dd, 1 H, J ) 8.0,8.0 Hz), 6.25 (m,
2 H), 6.08 (dd, 1H, J ) 8.0 Hz). 13NMR (DMSO-d6) 168.5,
162.1, 147.5, 137.8, 137.1, 136.6, 133.0, 128.9, 127.6, 127.5,
126.9, 122.8, 120.9, 117.7, 116.1, 115.8, 113.0, 111.3. Anal.
Calcd for C18H12Cl2N2O4 0.13 H2O: C, 54.93; H, 3.14; N,
7.12. Found: C, 54.88; H, 3.51; N, 0.6.95. Karl Fischer
Analysis: 0.6% H2O. MS m/z: (M+) calcd 390.0, obsd (M
+ H) 391.1.
Purification of a Sample Containing des-Chloro Im-
purity. A mixture of 517 g (E)-4,6-dichloro-3-[2-(3-ami-
nophenyl)-2-carboxyethenyl]-1H-indole-2-carboxylic acid
(5.8% des-chloro impurity, 5.4% Z isomer, ∼10% H2O) and
12.5 L of methanol was charged to a 22-L three-necked flask
fitted with a stirrer, thermometer, condenser, and a continu-
ous N2 purge. The mixture was heated to and held at reflux
for 30 min. The resulting turbid solution was clarified through
dicalite washing with 1 L of methanol. The filtrate was
concentrated at atmospheric pressure to a volume of ∼5 L.
The resulting slurry was cooled to 25 °C and filtered. The
filter cake was washed with 2 × 300 mL of methanol. A
recovery of 307.1 g was obtained at this point with an HPLC
analysis of 5.8% des-chloro impurity with essentially no Z
isomer. The 307.1 g of (E)-4,6-dichloro-3-[2-(3-aminophen-
yl)-2-carboxyethenyl]-1H-indole-2-carboxylic acid and 3.5
L of methanol was charged to a 12-L three-necked flask fitted
with a stirrer, thermometer, condenser, and a continuous N2
purge. The mixture was heated to and held at reflux for 18
h. The slurry was cooled to 25 °C and filtered, washing with
2 × 300 mL of methanol. The (E)-4,6-dichloro-3-[2-(3-
aminophenyl)-2-carboxyethenyl]-1H-indole-2-carboxylic acid
was dried at 45 °C/50 Torr. A yield of 241.3 g, 47%
recovery, was obtained with an HPLC analysis of 0.9% des-
chloro impurity with essentially no Z isomer.
A mixture of 2.66 kg (5.68 mol) of (E)-4,6-dichloro-3-
[2-carboxy-2-[3-[(1-methylethylidene)amino]ethenyl]-1H-in-
dole-2-carboxylic acid monohydrochloride, 10 and 26.6 L
of H2O was charged to a 35-L glass flask fitted with a stirrer.
The mixture was stirred for 20 min. The pH was adjusted
from 1.3 to 6.8 over 30 min using 3.1 L of 3 M NaOH. The
resulting solution was clarified through sintered glass using
1 L of H2O to wash. The pH of the filtrate was adjusted to
3.1 using 3.0 L of 3 M HCl. The resulting slurry was stirred
Received for review May 19, 2000.
OP000286S
Vol. 4, No. 6, 2000 / Organic Process Research & Development
•
487