A practical synthesis of multitargeted antifolate LY231514

Article abstract of DOI:10.1021/op9802172

A concise and scalable synthesis of LY231514 (1), a new pyrrolo[2,3-d]pyrimidine-based antitumor agent, is presented. Reaction of 2-bromo-4-arylbutanal 9 with 2,4-diamino-6-hydroxypyrimidinc (10) regioselectively provided pyrrolo[2,3-d]pyrimidine 11, representing the core structure of the drug, in good yield. Assimilation of the glutamic acid residue by conventional means completed the synthesis. Development of the optimized synthetic route emphasized avoiding isolation of the relatively unstable aldehyde and bromoaldehyde intermediates.

Full text of DOI:10.1021/op9802172

Organic Process Research & Development 1999, 3, 184−188
A Practical Synthesis of Multitargeted Antifolate LY231514
Charles J. Barnett,* Thomas M. Wilson, and Michael E. Kobierski
Chemical Process Research and DeVelopment DiVision, Lilly Research Laboratories, a diVision of Eli Lilly and Company,
Indianapolis, Indiana 46285-4813
Abstract:
A concise and scalable synthesis of LY231514 (1), a new pyrrolo-
[2,3-d]pyrimidine-based antitumor agent, is presented. Reaction
of 2-bromo-4-arylbutanal 9 with 2,4-diamino-6-hydroxypyri-
midine (10) regioselectively provided pyrrolo[2,3-d]pyrimidine
1
1, representing the core structure of the drug, in good yield.
Figure 1.
Taylor’s initial synthesis of LY231514 focused on the
Assimilation of the glutamic acid residue by conventional means
completed the synthesis. Development of the optimized synthetic
route emphasized avoiding isolation of the relatively unstable
aldehyde and bromoaldehyde intermediates.
construction of the two-carbon linker between the pyrrol-
2
opyrimidine and aryl glutamate moieties. Sequential pal-
ladium(0) coupling of acetylene (initially protected as the
TMS derivative) to aromatic (2) and heterocyclic (4) halides,
corresponding to the desired fragments as indicated in
Scheme 1, provided (via 3) unsymmetrical acetylene inter-
mediate 5. Conversion of 5 to 1 by standard methods of
catalytic reduction and hydrolytic removal of the ester and
amine protecting groups completed the synthesis. This
conceptually elegant and useful synthesis of this series of
antitumor agents lent itself readily to the preparation of 1
and its analogues for SAR studies.
Introduction
The concept that inhibitors of folate-requiring enzymes
could arrest the proliferation of malignant cells is perhaps
one of the oldest paradigms of cancer chemotherapy and
resulted in the discovery of methotrexate, one of the first
1
mechanism-based cancer drugs, more than 50 years ago.
Since that time, substantial research efforts have led to better
understanding of folate biochemistry and the mechanism of
methotrexate cytotoxicity, thus stimulating efforts to prepare
inhibitors of folate-requiring enzymes with improved clinical
utility in the treatment of cancer. The pyrrolo[2,3-d]pyrimi-
dine-based folate analogue, LY231514 (1, Figure 1), first
prepared by Taylor and co-workers at Princeton University
and evaluated for antitumor activity at Lilly, represents a
new generation of folate-requiring enzyme inhibitors with
This flexibility came at the price, however, of requiring
four steps to construct the ethylidene linker (two palladium-
mediated acetylene coupling operations, desilylation, and
catalytic reduction), in addition to the synthesis of the aryl
5
and heterocyclic coupling partners. The required 5-iodopy-
rrolo[2,3-d]pyrimidine intermediate proved difficult to pre-
pare. Direct iodination of 2-pivaloylamino-4-hydroxypyri-
midine with N-iodosuccinimide had given mixtures of 5- and
2
improved antitumor potency and spectrum. LY231514 has
6
-iodinated products as well as the 5,6-diiodide and starting
recently been found to possess inhibitory activity against
dihydrofolate reductase and C-1 synthase in addition to its
primary enzymatic target, thymidylate synthase, and has thus
2
material in the Taylor laboratory, and our experience with
this reaction was the same. For good results, it was necessary
to deliberately 5,6-diiodinate the pyrrolopyrimidine inter-
mediate and remove the 6-iodo substituent by selective
reduction with zinc in acetic acid. The zinc reduction required
careful control of stoichiometry and conditions to avoid
under- or overreduction (deiodination of the 5-position).
These requirements proved difficult to provide as the scale
increased. We developed an improved procedure for the
preparation of 4 involving in situ N,O-bis-silylation of the
pyrrolopyrimidine precursor and regiospecific C-5 monoio-
3
been designated as a multitargeted antifolate (MTA). Phase
II clinical trials have revealed clinically significant activity
against colorectal, lung, and breast cancer and promising
activity against a number of other tumor types. We describe
herein the development of a practical synthesis of LY231514
which is more suitable for support of advanced drug
development efforts.4
*
To whom correspondence should be addressed. Fax: (317) 276-6895.
E-mail: Barnett_Charles_J@lilly.com.
(
1) For recent overviews of antifolate research related to antitumor therapy,
see: Berman E. M.; Werbel, L. M. J. Med. Chem. 1991, 34, 479 and
references therein. Taylor, E. C. J. Heterocycl. Chem. 1990, 27, 1-12 and
references therein.
(4) A preliminary report of this work has been published: Barnett, C. J.; Wilson,
T. M.; Kobierski, M. E. In Chemistry and Biology of Pteridines and Folates
1997; Pfleiderer, W., Rokos, H., Eds.; Blackwell Wissenschafts-Verlag:
Berlin, 1997; pp 123-126.
(
2) Taylor, E. C.; Kuhnt, D.; Shih, C.; Rinzel S. M.; Grindey, G. B.; Barredo,
J.; Jannatipour, M.; Moran, R. G. J. Med. Chem. 1992, 35, 4450-4454.
3) Shih, C.; Chen, V. J.; Gossett, L. S.; Gates, S. B.; MacKellar, W. C.; Habeck,
L. L.; Shackelford, K. A.; Mendelsohn, L. G.; Soose, D. J.; Patel, V. F.;
Andis, S. L.; Bewley, J. R.; Rayl, E. A.; Moroson, B. A.; Beardsley, G. P.;
Kohler, W.; Ratnam, M.; Schultz, R. M. Cancer Res. 1997, 57, 1116-
(5) Intermediate 2 could be prepared from 4-iodobenzoyl chloride and dimethyl-
L-glutamate but is not commercially available. Preparation of 4 proceeded
via (commercially unavailable) 2-amino-4,7-dihydro-4-oxo-1H-pyrrolo[2,3-
d]pyrimidine (acylation with pivaloyl chloride, iodination, and zinc reduction
(
2
as described by Taylor et al. ). The required pyrrolo[2,3-d]pyrimidine starting
material could be prepared by literature methods from, for example, 2,4-
diamino-6-hydroxypyrimidine, as described in ref 7.
1
123.
1
84
•
Vol. 3, No. 3, 1999 / Organic Process Research & Development
10.1021/op9802172 CCC: $18.00 © 1999 American Chemical Society and Royal Society of Chemistry
Published on Web 04/30/1999

Scheme 1
It appears that the first examples of synthesis of pyrrolo-
[2,3-d]pyrimidines via cyclocondensation of a 2-amino-4-
pyrimidone with R-haloaldehydes were carried out by
7
Hitchings and Noell and Robins. These authors reported that
the reaction of chloroacetaldehyde with 2,4-diamino-6-
hydroxypyrimidine derivatives afforded 2-amino-4-hydroxy-
pyrrolo[2,3-d]pyrimidines. Secrist and Liu extended the
process to the synthesis of various 5- and 6-substituted
pyrrolo[2,3-d]pyrimidines by utilizing various R-halo ketones
8
and aldehydes. These workers established that fast alkylation
of the pyrimidine ring at C-5 by the halocarbonyl compound
tended to predominate over attack at other sites (O, N), thus
reducing the possibilities for formation of regioisomeric
products. Competing alkylation of the pyrimidone oxygen
was observed in the case of certain haloketone examples,
leading to mixtures containing both pyrrolo- and furo-fused
pyrimidine products, but a limited number of haloaldehyde
examples were reported to give only pyrrolo[2,3-d]pyrim-
idines. Reaction of 2,4-diamino-6-hydroxypyrimidine with
2
-chloropropionaldehyde in dimethyl sulfoxide (DMSO) in
the presence of a substoichiometric amount of sodium
bicarbonate gave, for example, 2-amino-4-hydroxy-5-meth-
ylpyrrolo[2,3-d]pyrimidine in 60% (crude) yield. We were
encouraged by the reported regioselectivity of the halopro-
pionaldehyde example; the regioisomeric 6-methyl substi-
tuted product was apparently not observed. Because of the
limited number of examples of this reaction reported here
or elsewhere, however, the effect that a more complex
haloaldehyde structure would have on the outcome of the
reaction was not clear.
6
dination of the resulting silylated intermediate. The latter
method represented a significant improvement, but the
experimental challenges and expense of the two palladium-
catalyzed coupling steps remained unremedied. The coupling
of acetylenic intermediates to the iodinated pyrrolopyrimidine
system was complicated by a strong tendency for the
iodoheterocycle to undergo reductive deiodination rather than
coupling, causing the yield of desired product to be moder-
ated.
Results and Discussion
Synthetic Strategy
Our synthetic scheme for utilization of the haloaldehyde
cyclocondensation process in the synthesis of LY231514 is
depicted in Scheme 2. Palladium(0) coupling of methyl
We considered that our reliance on palladium coupling
chemistry might be significantly reduced by focusing on the
formation of the pyrrolo[2,3-d]pyrimidine ring system as a
point of synthetic convergence, as indicated retrosynthetically
below. In particular, the problems involved in preparing the
4
-bromobenzoate with 3-butyn-1-ol by a modified literature
9
procedure gave crystalline 6, which was reduced over
palladium on carbon in methylene chloride. Filtration of the
catalyst afforded a methylene chloride solution of alcohol 7
5-iodopyrrolo[2,3-d]pyrimidine 4 and coupling it to the
(
in quantitative yield), which was utilized directly in a
arylacetylene fragment could be circumvented by this ap-
proach.
1
0
11
TEMPO -catalyzed sodium hypochlorite oxidation pro-
1
2
cedure, providing the known aldehyde 8, which was not
isolated. Addition of 5,5-dibromobarbituric acid13 (DBBA)
and a small amount of HBr in acetic acid to the methylene
chloride solution of 8 effected conversion to the R-bromoal-
dehyde 9. Choice of the dibromobarbituric acid reagent was
made on the basis of its selectivity toward R-bromination14
and the convenience of manipulation of a stable, crystalline
brominating reagent. Filtration of the reaction mixture
1
5
removed byproduct barbituric acid, and any remaining
brominating agent was removed by washing the solution with
(
7) Hitchings, G. H. British Patent 812,366, 1959; Chem. Abstr. 1960, 54, 592i.
Noell, C. W.; Robins, R. K. J. Heterocycl. Chem. 1964, 1, 34-41.
8) Secrist, J. A.; Liu, P. S. J. Org. Chem. 1978, 43, 3937-3941.
9) Taylor, E. C.; Harrington, P. M. J. Org. Chem. 1990, 55, 3222-3227.
(
(
(10) 2,2,6,6-Tetramethyl-1-piperidinyloxy free radical, Aldrich Chemical Co.
(11) Anelli, P. L.; Montanari, F.; Quici, S. Org. Synth. 1990, 69, 212-219.
(12) This substance has previously been prepared by the oxidation of 7 with
pyridinium chlorochromate (ref 9).
(
6) Barnett, C. J.; Kobierski, M. E. J. Heterocycl. Chem. 1994, 31, 1181-
183.
1
(13) Grundke, G.; Keese, K.; Rimpler, M. Chem. Ber. 1985, 118, 4288-4291.
Vol. 3, No. 3, 1999 / Organic Process Research & Development 185
•

Scheme 2
sodium thiosulfate. After aqueous workup, the solution was
concentrated and diluted with acetonitrile to effect solvent
exchange. Addition of commercially available 2,4-diamino-
chloride was successfully utilized. As in all DMSO based
oxidation methods, however, removal of the byproduct
dimethyl sulfide was difficult unless the aldehyde was
isolated and purified by, for example, silica chromatography.
When crude 8 was taken to the bromination as a methylene
chloride solution (from the workup of the oxidation), a
significant amount of dimethyl sulfide was carried into the
bromination, consuming the brominating reagent. The dim-
ethyl sulfide could be destroyed by addition of excess
brominating agent, but the calculation of the required excess
and related analytical work proved to be inconvenient. The
TEMPO oxidation provided a solution to this problem by
eliminating sulfide residue carryover.
6-hydroxypyrimidine and aqueous sodium acetate and heat-
ing to 45 °C caused cyclocondensation and precipitation of
pyrrolo[2,3-d]pyrimidine 11 from the reaction mixture in
67% overall yield based on 7. The yield and quality of the
key intermediate 11 from the rather complex sequence
beginning with arylbutanol 7 was quite remarkable. Despite
numerous formal possibilities for the formation of other
bicyclic products, there was no indication of the presence
of isomeric cyclization products in the NMR spectra of crude
1
1.
The four-step processing sequence for synthesis of 11
beginning with 6) was designed to avoid the isolation of
oily intermediates 7, 8, and 9, the latter two of which were
Saponification of 11 with aqueous sodium hydroxide
followed by acidification afforded the carboxylic acid
derivative 12, which was elaborated to 13 by means of the
(
16
unstable toward storage conditions as neat liquids. Selection
of the conditions for oxidation of alcohol 7 to the corre-
sponding aldehyde 8 proved to be an important consideration.
An apparent variant17 of the Pfitzner-Moffatt oxidation of
alcohols involving dimethyl sulfoxide-phosphorus pentoxide
in the presence of triethylamine as the oxidant in methylene
18
chlorodimethoxytriazine active ester peptide coupling
method. Reaction of 12 with 2-chloro-4,6-dimethoxy-1,3,5-
triazine (CDMT) in the presence of N-methylmorpholine in
DMF solution followed by reaction of the resulting dimethoxy-
s-triazinyl ester with diethyl L-glutamate afforded crude 13,
which was isolated via crystallization as the p-toluene-
sulfonate salt (14). It is noteworthy that the 2-amino group
in 12 did not require protection during the coupling reaction.
Neutralization of the salt and saponification of 13 with
aqueous sodium hydroxide followed by acidification with
hydrochloric acid gave LY231514 (1) as the free diacid,
which was crystallized as the disodium salt (15) for
pharmaceutical use. Careful analytical studies showed that
no racemization of the glutamate center occurred in either
the coupling or saponification steps.
(
14) Oxidation of aldehydes to the corresponding acyl halides is a potential
problem in the bromination of aldehydes with elemental bromine, particularly
if the reaction mixture is contaminated with water. For a discussion of
aldehyde bromination, see: DeKimpe, N.; Verhe, R. In The Chemistry of
R-Haloketones, R-Haloaldehydes, and R-Haloimines; Patai, S., Rappoport,
Z., Eds.; Updates from the Chemistry of Functional Groups; Wiley-
Interscience: New York, 1988; Chapter 3. We found N-bromosuccinimide
and related reagents to be unsuitable for this transformation due to low
yields and competing aldehyde oxidation.
(15) The brominating reagent can be conveniently regenerated by bromination
2
of the barbituric acid by bromine. Two moles of Br is required per mole
of barbituric acid, but both bromine atoms in the reagent are utilized in the
bromination, so the overall stoichiometry is the same as that with bromine
itself. See ref 13.
(
16) Aldehyde 8 was found to be reasonably stable as an oil at room temperature
when chromatographically pure, but the crude material tended to form acetal
polymers on standing which were not easily cracked to recover the aldehyde.
Bromoaldehyde 9 was unstable toward storage in neat form at refrigerator
freezer temperature. It could be kept as a refrigerated (5 °C) methylene
chloride solution for 24 h without significant decomposition.
(17) Taber, D. F.; Amedio, J. C.; Jung, K.-Y. J. Org. Chem. 1987, 52, 5621-
2 5 3
5622. The DMSO-P O -Et N method was attractive for large-scale
application because the process could be carried out at or near room
temperature. The oxalyl chloride-based Swern procedure requires much
lower temperatures.
(18) Kaminski, Z. J. Tetrahedron Lett. 1985, 26, 2901-2904.
186
•
Vol. 3, No. 3, 1999 / Organic Process Research & Development

The synthesis of LY231514 described herein illustrates
the power of retrosynthetic disconnections based on hetero-
cyclic ring synthesis, in this case the pyrrolo[2,3-d]pyrimidine
system, for providing advantageously convergent and ex-
perimentally simple routes to complex synthetic targets. The
cyclocondensation of aminopyrimidone 10 with its R-bro-
moaldehyde partner 9 was the basis for a practical and
scalable synthesis of LY231514 (1).
To the resulting solution, cooled to -10 to 0 °C, was added
0.045 kg of TEMPO catalyst and a solution of 1.72 kg of
potassium bromide in 5.1 L of water. A total of 153 L of
12% aqueous sodium hypochlorite solution (adjusted to pH
9.5 by addition of sodium bicarbonate) was next added at a
rate such that the temperature of the reaction mixture did
not exceed 20 °C. Complete conversion of starting material
was confirmed by HPLC analysis. The layers were separated,
and the methylene chloride phase was washed sequentially
with a solution of 0.095 kg of potassium iodide in 70 L of
1 N HCl, 70 L of 1 N sodium thiosulfate, 70 L of 5% sodium
bicarbonate, and 70 L of half-saturated brine solution. The
Experimental Section19
4-(4-Hydroxy-1-butynyl)benzoic Acid Methyl Ester (6).
To a mixture of 57.6 kg (268 mol) of methyl 4-bromoben-
zoate and 420 L of ethyl acetate were added 310 g (1.7 mol)
of palladium(II) chloride, 840 g (3.2 mol) of triphenylphos-
phine, 610 g (3.2 mol) of cuprous iodide, 64 kg (880 mol)
of diethylamine, and 23.6 kg (337 mol) of 3-butyn-1-ol. The
resulting mixture was heated to 50 °C for about 4 h. The
mixture was cooled and filtered, and the filtered solids
2
1
resulting methylene chloride solution of 8 was taken to
the next step without isolation. A sample obtained by
concentration of an aliquot from a smaller scale run had
1
9
H NMR data identical to the reported data. The semicar-
bazone derivative had mp 149-151 °C. Anal. Calcd for
: C, 59.30; H, 6.51; N, 15.96. Found: C, 59.25;
H, 6.47; N, 15.67.
RS)-4-(3-Bromo-4-oxobutyl)benzoic Acid Methyl Es-
17 13 3 3
C H N O
(diethylamine hydrobromide and residual catalyst) were
washed with 180 L of ethyl acetate. The filtrate was washed
with a solution of 85 kg of sodium bisulfate in 310 L of
deionized water and then separated. The organic layer was
washed with saturated sodium chloride. Sodium sulfate (42
(
ter (9). The methylene chloride solution of 8 from the
previous step was added to a mixture of 24.5 kg (85.8 mol,
13
1
1
71.6 equiv) of 5,5-dibromobarbituric acid (DBBA) and
.4 L of 30% hydrobromic acid in acetic acid in 150 L of
20
kg), silica gel (42 kg), and 18 kg (wet wt) of Deloxan THP
resin were added, and the resulting slurry was stirred for
about 20 min. The mixture was filtered, and the filter cake
was washed with 300 L of ethyl acetate. The filtrate was
concentrated under vacuum to 190 L. Ethyl acetate (105 L)
was added, followed by 435 L of heptane, causing the
product to crystallize. The mixture was cooled to 0-5 °C
and stirred for about 2 h. The crystals were filtered, washed
methylene chloride. The mixture was stirred at 25 °C until
complete conversion of starting material was indicated by
HPLC analysis. The mixture was filtered to remove barbituric
acid formed during the bromination. The filtrate was washed
with 260 L of 1 N sodium thiosulfate, and the layers were
separated. The organic layer was washed with 260 L of 5%
sodium bicarbonate solution and 225 L of half-saturated brine
solution and then concentrated under vacuum to 90 L. The
solution was diluted with 400 L of acetonitrile, and residual
methylene chloride was removed from the mixture by
distillation, reducing the volume to 255 L. The resulting
with 200 L of cold heptane (0-5 °C), and dried, affording
9
4
9
5.6 kg (83.4% yield) of 6, mp 97.5-99 °C (lit. mp 95.5-
6 °C).
4-(4-Hydroxybutyl)benzoic Acid Methyl Ester (7). A
1.0-kg portion of 5% palladium on carbon catalyst was
2
2
solution of compound 9 in acetonitrile was taken to the
slurried in 25 L of methylene chloride. The slurry was
charged to a nitrogen-purged 30-gal autoclave with an
additional 45 L of methylene chloride. Compound 6, 9.82
kg (48.1 mol), was added with 15 L of methylene chloride.
The resulting mixture was stirred under 50 psig of hydrogen
for 3.5 h. The catalyst was filtered and washed with 20 L of
methylene chloride. The resulting methylene chloride solu-
tion was found to contain 9.94 kg of 7 (99.2% based on 6)
by analysis of the mass of the residue from evaporation of
an aliquot. The solution of 7 was carried directly into the
next step. A small sample was concentrated to dryness in
vacuo and identified by NMR spectral comparison with an
authentic sample made by the method of Taylor and
Harrington.9
next step without isolation. An analytical sample was
prepared separately by flash chromatography. H NMR
1
(CDCl
3
): δ 2.22 (m, 1H), 2.38 (m, 1H), 2.80 (m, 1H), 2.94
(
2
m, 1H), 3.91 (s, 3H), 4.17 (m, 1H), 7.28 (d, J ) 8.2 Hz,
13
H), 7.98 (d, J ) 8.2 Hz, 2H), 9.47 (d, J ) 2.2 Hz, 1H). C
): δ 32.6, 32.7, 52.0, 54.3, 128.5, 130.0, 145.2,
66.8, 192.3. IR (CHCl ): 2955, 1720, 1611, 1437, 1284,
180, 1114. MS (FD): m/z 286 (100), 284 (98). Anal. Calcd
: C, 50.55; H, 4.60. Found: C, 50.31; H,
NMR (CDCl
1
1
3
3
for C12H13BrO
3
4
.72.
-[2-(2-Amino-4,7-dihydro-4-oxo-1H-pyrrolo[2,3-d]py-
4
rimidin-5-yl)ethyl]benzoic Acid Methyl Ester (11). To the
acetonitrile solution of compound 9 from the preceding step
was added 255 kg of deionized water, 16.2 kg (128.6 mol)
of 2,4-diamino-6-hydroxypyrimidine (10), and 21.1 kg (256.9
mol) of sodium acetate. The slurry was heated to 40 °C for
4
-(4-Oxobutyl)benzoic Acid Methyl Ester (8). A solu-
tion of 29.0 kg (139 mol) of 7 in 345 L of methylene chloride
prepared in three sections as described above) was reduced
to a volume of 155 L by partial distillation of the solvent.
(
(21) Accelerated rate calorimetry data for 8 as a neat liquid: no exothermic
events until slow self-heating (0.04 °C/min) was encountered at 287 °C.
(
19) Proton NMR spectra were obtained at 300 MHz. Carbon-13 NMR spectra
were obtained at 75 MHz. NMR peak positions are reported in ppm
downfield from internal tetramethylsilane (δ). Other spectral data were
obtained using standard instrumentation.
(22) Accelerated rate calorimetry data for 9 as a neat liquid: exothermic
decomposition at 85 °C and maximum self-heat rate of 25.5 °C/min at 125
°C. Solution of 9, 14% in acetonitrile: onset of exotherm at 100 °C with
maximum self-heat rate of 0.44 °C at 116 °C. Maximum reaction temperature
for conversion of 9 to 11 was limited to 50 °C to provide a 50 °C margin
of safety.
(20) Degussa AG, Frankfurt, Germany. The resin was used to remove residual
palladium.
Vol. 3, No. 3, 1999 / Organic Process Research & Development
•
187

about 3 h. The mixture was cooled to 20 °C and filtered.
6.04; N, 14.48. Found: C, 59.40; H, 6.08; N, 14.41.
The crystals were washed with 200 L of acetonitrile/water
N-[4-[2-(2-Amino-4,7-dihydro-4-oxo-1H-pyrrolo[2,3-d]-
pyrimidin-5-yl)ethyl]benzoyl]-L-glutamic Acid Diethyl
Ester 4-Methylbenzenesulfonic Acid Salt (14). To the
resulting solution of 13 was added a solution of 6.12 kg (32.2
mol) of p-toluenesulfonic acid monohydrate in 105 L of
ethanol, and the resulting suspension was heated under reflux
for about 3 h. The mixture was cooled to about 25 °C with
stirring. The crystals of 14 were filtered and washed with
85 L of ethanol. The wet cake was reslurried in 155 L of
ethanol, refluxed for 1 h, and cooled to 22 °C. The crystals
were filtered, washed with 85 L of ethanol, and dried to
provide 7.42 kg of 14. A total of 13.6 kg of 14 thus obtained
from separate runs was combined and suspended in 57 L of
DMSO. The mixture was heated to 80 °C and stirred for 1.5
h. Hot ethanol (230 L at 75 °C) was added to the solution,
which then was cooled to 5-10 °C. The resulting mixture
was stirred for 30 min and filtered. The wet cake was
reslurried with 175 L of ethanol (5-10 °C), filtered, and
washed with 60 L of ethanol. The crystals were dried to
provide 11.2 kg (72.4% from 12) of purified 14, mp 255-
(
1:1) and dried under vacuum, affording 29.2 kg (67% yield
1
based on 7) of 11, mp > 250 °C. H NMR (DMSO-d
.88 (m, 4H), 3.78 (s, 3H), 5.97 (s, 2H), 6.26 (d, J ) 1.6
Hz, 1H), 7.29 (d, J ) 8.1 Hz, 2H), 7.81 (d, J ) 8.2 Hz,
6
): δ
2
2
d
1
H), 10.1 (bd s, 1H), 10.6 (bd s, 1 H). 13C NMR (DMSO-
): δ 27.4, 35.8, 51.5, 98.3, 113.0, 117.1, 126.6, 128.2,
28.7, 147.9, 150.9, 151.8, 158.9, 165.8. UV (EtOH): λ 225
6
(
ꢀ 22 450), 243 nm (ꢀ 20 800). Anal. Calcd for C16
C, 61.53; H, 5.16; N, 17.94. Found: C, 61.42; H, 5.19; N,
7.81.
-[2-(2-Amino-4,7-dihydro-4-oxo-1H-pyrrolo[2,3-d]py-
15 4 3
H N O :
1
4
rimidin-5-yl)ethyl]benzoic Acid (12). Compound 11 (4.76
kg, 15.2 mol) was suspended in 54 L of 2 N aqueous sodium
hydroxide solution and heated to 40 °C for about 1.5 h, at
which time HPLC analysis showed the reaction to be
complete. The reaction mixture was diluted with 81 L of
ethanol and cooled to 20-25 °C. The pH was adjusted to
4.4 with 6 N aqueous hydrochloric acid solution, and the
resultant precipitate was filtered, washed with 57 L of a 1:1
solution of water and ethanol, and dried to provide 4.12 kg
37 5 9
260 °C. Anal. Calcd for C31H N O S: C, 56.78; H, 5.69;
N, 10.68. Found: C, 56.96; H, 5.67; N, 10.79.
1
(91%) of compound 12, mp > 250 °C. H NMR (DMSO-
d
6
): δ 2.86 (m, 4H), 6.11 (s, 2H), 6.28 (s, 1H), 7.26 (d, J )
N-[4-[2-(2-Amino-4,7-dihydro-4-oxo-1H-pyrrolo[2,3-d]-
pyrimidin-5-yl)ethyl]benzoyl]-L-glutamic Acid (LY231514,
1). Compound 14 (5.5 kg, 8.4 mol) was added to 38 L of 1
N aqueous sodium hydroxide. The solution was stirred under
nitrogen at about 25 °C until completion of the reaction was
indicated by HPLC analysis. The reaction mixture was
diluted with 43 L of ethanol and adjusted to pH 3 with
aqueous hydrochloric acid, causing the 1 free acid to
precipitate. The resulting slurry was heated with stirring to
70-75 °C and then cooled to 20-25 °C. The crystals were
filtered and washed with 50 L of ethanol-water (1:1). The
wet cake was reslurried in 54 L of ethanol-water (1:1),
filtered, and dried to yield 3.6 kg (100%) of 1, mp > 250
°C, identical in all respects (NMR, IR, UV, MS, HPLC) with
an authentic sample prepared by the method of Taylor et
8.1 Hz, 2H), 7.79 (d, J ) 8.1 Hz, 2H), 10.24 (s, 1H), 10.64
13
(
3
1
s, 1H), 12.40 (bd s, 1H). C NMR (DMSO-d
6
): δ 27.3,
5.7, 98.2, 113.0, 117.1, 127.6, 127.9, 128.7, 147.2, 150.2,
51.6, 158.7, 166.8. UV (EtOH): λ 225 nm (ꢀ 24 000). Anal.
: C, 60.40; H, 4.73; N, 18.78. Found:
14 4 3
Calcd for C15H N O
C, 60.59; H, 4.77; N, 18.64.
N-[4-[2-(2-Amino-4,7-dihydro-4-oxo-1H-pyrrolo[2,3-d]-
pyrimidin-5-yl)ethyl]benzoyl]-L-glutamic Acid Diethyl
Ester (13). To a suspension of 3.84 kg (12.9 mol) of
compound 12 in 22 L of DMF was added 4.04 kg (39.9 mol)
of N-methylmorpholine, followed by the addition of 3.08
kg (17.5 mol) of 2-chloro-4,6-dimethoxy-1,3,5-triazine (CD-
MT). The resulting mixture was stirred for 1.5 h at 25 °C,
at which time HPLC analysis showed that the reaction was
complete. L-Glutamic acid diethyl ester hydrochloride (3.40
kg, 14.2 mol) was added, and stirring was continued at 25
1
8
2
al.
N-[4-[2-(2-Amino-4,7-dihydro-4-oxo-1H-pyrrolo[2,3-d]-
pyrimidin-5-yl)ethyl]benzoyl]-L-glutamic Acid Disodium
Salt (15). The disodium salt was obtained by neutralization
°
C until complete conversion of 12 was determined by
HPLC. To the reaction mixture were added 71 L of
methylene chloride and 71 L of deionized water, and the
mixture was stirred for 15 min. The layers were separated,
and the lower (organic) layer was concentrated by atmo-
spheric pressure distillation to 30 L and then diluted with
with sodium hydroxide and crystallization from water-
1
ethanol as a hygroscopic solid in 85% yield. H NMR (CD
3
-
OD): δ 2.25 (m, 4H), 2.95 (m, 4H) 3.28 (dd, J ) 1.6, 3.2
Hz, 1H), 6.26 (s, 1H), 7.20 (d, J ) 8.2 Hz, 2H), 7.73 (d, J
) 8.2 Hz, 2H). Chiral HPLC analysis indicated that the
enantiomeric purity of the product was >99.5%.
5
5 L of ethanol. The solution containing 13 was taken
directly to the next step. An analytical sample was prepared
separately by chromatography on silica gel, mp 169-171
1
°
C. H NMR (DMSO-d
6
): δ 1.14 (t, 3H), 1.17 (t, 3H), 2.03
Acknowledgment
(
m, 2H), 2.42 (t, J ) 7.4 Hz, 3H), 2.89 (m, 4H), 4.02 (q,
The authors gratefully acknowledge the engineering
contributions of Jason Cronin, Robert Towsley, Brian
Southern, and their associates to the successful implementa-
tion of this chemistry on a pilot plant scale.
2
1
2
H), 4.08 (q, 2H), 4.41 (m, 1H), 6.00 (s, 2H), 6.29 (d, J )
.7 Hz, 1H), 7.27 (d, J ) 8.2 Hz, 2H), 7.76 (d, J ) 8.2 Hz,
H), 8.82 (d, J ) 7.4 Hz, 1H), 10.15 (s, 1H), 10.60 (s, 1H).
1
3
C NMR (DMSO-d ): δ 14.0, 25.7, 27.9, 30.2, 36.1, 51.9,
6
5
1
9.9, 60.5, 98.7, 113.4, 117.6, 127.3, 128.1, 131.1, 146.6,
51.3, 152.2, 159.3, 166.6, 171.8, 172.2. UV (EtOH): λ 224
Received for review October 30, 1998.
OP9802172
29 5 6
nm (ꢀ 26 240). Anal. Calcd for C24H N O : C, 59.62; H,
188
•
Vol. 3, No. 3, 1999 / Organic Process Research & Development