Highly chemoselective trichloroacetimidate-mediated alkylation of ascomycin: A convergent, practical synthesis of the immunosuppressant L- 733,725

Article abstract of DOI:10.1021/jo981805g

L-733,725, a new immunosuppressant drug candidate, was prepared by a highly chemoselective alkylation of the macrolide ascomycin at the C32 hydroxy position with the imidazolyl trichloroacetimidate 16. The trichloroacetimidate-activated side chain 16 was prepared by an efficient fourstep sequence in 42% overall yield. The high chemoselectivity in the alkylation of the C32 hydroxy group of the unprotected ascomycin was the result of the synergetic effects of the electron-donating protecting group on the imidazole 16, the polar, moderately basic solvent, and the strong acid catalyst. N,N-Dimethylpivalamide mixed with acetonitrile was found to be the best solvent and trifluromethanesulfonic acid the best catalyst. This synthesis coupled with a resin column purification of L-733,725 followed by crystallization of its tartrate salt has been used to make multikilogram quantities of the bulk drug with consistent and high purity.

Full text of DOI:10.1021/jo981805g

J . Org. Chem. 1999, 64, 1859-1867
1859
High ly Ch em oselective Tr ich lor oa cetim id a te-Med ia ted Alk yla tion
of Ascom ycin : A Con ver gen t, P r a ctica l Syn th esis of th e
Im m u n osu p p r essa n t L-733,725
Zhiguo Song,* Anthony DeMarco, Mangzhu Zhao, Edward G. Corley, Andrew S. Thompson,
J ames McNamara, Yulan Li, Dale Rieger, Paul Sohar, David J . Mathre, David M. Tschaen,
Robert A. Reamer, Martha F. Huntington, Guo-J ie Ho, Fuh-Rong Tsay, Khateeta Emerson,
Richard Shuman, Edward J . J . Grabowski, and Paul J . Reider
Department of Process Research, Merck Research Laboratories, P.O. Box 2000,
Rahway, New J ersey 07065
Received September 4, 1998
L-733,725, a new immunosuppressant drug candidate, was prepared by a highly chemoselective
alkylation of the macrolide ascomycin at the C32 hydroxy position with the imidazolyl trichloro-
acetimidate 16. The trichloroacetimidate-activated side chain 16 was prepared by an efficient four-
step sequence in 42% overall yield. The high chemoselectivity in the alkylation of the C32 hydroxy
group of the unprotected ascomycin was the result of the synergetic effects of the electron-donating
protecting group on the imidazole 16, the polar, moderately basic solvent, and the strong acid
catalyst. N,N-Dimethylpivalamide mixed with acetonitrile was found to be the best solvent and
trifluromethanesulfonic acid the best catalyst. This synthesis coupled with a resin column
purification of L-733,725 followed by crystallization of its tartrate salt has been used to make multi-
kilogram quantities of the bulk drug with consistent and high purity.
FK-506 and related compounds are immunosuppres-
sants that have been under development for treatment
of rejection of transplanted organs and for psoriasis.1
Although FK-506 has been approved for clinical use, its
toxicity still makes it desirable to find a more potent and
less toxic drug. L-733,725 is potentially one such com-
pound under development.1a,b The structure of L-733,725,
as shown in Scheme 1, consists of two parts, the asco-
mycin (i.e., L-683,590 or FK-520) macrocycle and an aryl
imidazole side chain connected via an ether linkage on
C32 of the macrocycle. The synthesis of the macrocycle
by chemical or microbial methods is well established, and
for the purpose of large-scale production, fermentation
is the method of choice.2 The principal challenge in
designing a practical synthesis of L-733,725 is focused
upon a suitably activated imidazole side chain and
selectively coupling it to the C32 hydroxy group of the
macrocycle. A retrosynthetic analysis is shown in Scheme
1. The key steps are the imidazole formation and the
ether linkage formation. Two routes to the imidazole
portion of the molecule were envisioned. The first one is
from the condensation of the aryl glyoxal 1 (hydrate),
ammonia, and the aldehyde. The other is via palladium-
or nickel-mediated coupling of the aryl halide and the
metalated imidazole. Manipulations involving the mac-
rocycle are to be minimized. This convergent route should
be more economical than the existing route, where the
side chain was built on the macrocycle in a multistep
linear fashion.1a,b
Resu lts a n d Discu ssion
Im id a zole Syn th esis. One of the classical methods
of imidazole synthesis is condensation of glyoxal, am-
monia, and an aldehyde unit.3a For the preparation of
the aryl glyoxal 1 several methods were investigated.4
Oxidation of the methyl aryl ketone 2 (ArCOMe) by
DMSO in the presence of aqueous HBr was chosen and
run on a pilot plant scale. The reaction was carried out
at 60 °C, and the side product, dimethyl sulfide, was
removed with a sweep of nitrogen. The glyoxal monohy-
drate 1 was crystallized from aqueous DMSO in 77%
yield.
Making the imidazole from the aryl glyoxal, ammonia,
and an aldyhyde unit proved nontrivial.3a As shown in
Scheme 2, when glycolaldehyde dimer 4 was used, a 2/1
mixture of the desired imidazole 5 and the imidazole 6
without the hydroxymethyl moiety was produced. Be-
cause of the problem in selectively protecting the imid-
(1) (a) Goulet, M.; McAlpine, S. R.; Staruch, M. J .; Koprak, S.;
Dumont, F. J .; Cryan, J . G.; Wiederrech, G. J .; Rosa, R.; Wilusz, M.
B.; Peterson, L. B.; Wyvratt, M. J .; Parsons, W. H. Bioorg. Med. Chem.
Lett. 1998, 8, 2253. (b) Goulet, M.; Sinclair, P. J .; Wong, F.; Wyvratt,
M. J . US Patent 5247076, 1993. (c) Kawai, M.; Gunawardana, I. W.
K.; Mollison, K. W.; Hsieh, G. C.; Lane, B. C.; Luly, J . R. Biiorg. Med.
Chem. Lett. 1998, 8, 935. (d) Wagner, R.; Rhodes, T. A.; Ore, Y. S.;
Lane, B. C.; Hsieh, G.; Mollison, K. W.; Luly, J . R. J . Med. Chem. 1998,
41, 1764. (e) Goulet, M. T.; Rupprecht, K. M.; Sinclair, P. J .; Wyvratt,
M. J .; Parsons, W. H. Perspect. Drug Discovery Des. 1994, 2, 145. (f)
Van Duyne, G. D.; Standaert, R. F.; Karplus, P. A.; Schreiber, S. L.;
Clardy, J . Science 1991, 252, 839.
(2) (a) J ones. T.; Reamer, R. A., Desmond, R.; Mills. S. G. J . Am.
Chem. Soc. 1990, 112, 2998. (b) Byrne, K. M., Shafiee, A.; Nielsen, J .
B., Arison, B.; Monaghan, R. L.; Kaplan, L. In Microbial Metabolites;
Nash, C., Hunter-Cevera, J ., Cooper, R., Eveleigh, D. E., Hamill, R.,
Eds.; Wm. C. Brown Publishers: Dubuque, IA, 1992; Vol. 32, p 29.
(3) (a) Grimmett, M. R. In Comprehensive Heterocyclic Chemistry;
Katritzky, A. R., Rees, C. W., Potte, K. T., Eds.; Pergamon Press: New
York, 1984; Vol. 5, pp 457-497. (b) Sigehara, I.; Nakajima, T.;
Nishimura, S.; Ohshima, T. EP 365030A1, 1989. (c) Walia, J . S.; Walia,
A. S.; Lankin, D. C., Petterson, R. C.; Singh, J . J . Heterocycl. Chem.
1985, 22, 1117.
(4) (a) Mikol, G. J .; Russel, G. A. Organic Syntheses; Wiley: New
York, 1963; Collect. Vol. IV, p 937. (b) Riley, H. A.; Gray, A. R. Organic
Syntheses; Wiley: New York, 1943; Collect. Vol. II, p 509. (c) Desmond,
R.; Mills, S.; Volante. R. P., Shinkai, I. Synth. Commun. 1989, 19, 379.
(d) Keifanto, M. M.; Delepine, M. M. Compt. Rend. 1962, 254, 493. (e)
Floyd, M. B.; Du, M. T.; Fabio, P. F.; J acob, L. A.; J ohnson, B. D. J .
Org. Chem. 1985, 50, 5022.
10.1021/jo981805g CCC: $18.00 © 1999 American Chemical Society
Published on Web 02/25/1999

1860 J . Org. Chem., Vol. 64, No. 6, 1999
Song et al.
Sch em e 1
Sch em e 2
Sch em e 3
3 equiv of 10 to a solution of 3 equiv of ammonium acetate
and 1 equiv of 10 in 1-2 h. The aryl imidazole ester 11
was formed in ∼75% assay yield and isolated in 60-65%
yield by direct crystallization. The analogous phenyl
imidazole ester 11B (similar to 11 but Ar ) Ph) can also
be made with the same procedure in 58% yield.
P r otectin g Gr ou p Ch em istr y. Because of the insta-
bility of ascomycin toward strong bases, the ether linkage
between the side chain and the macrocycle has to be
made under nearly neutral or acidic conditions. Tri-
chloroacetimidate-mediated coupling is the most attrac-
tive convergent method for this reason.5 In addition, a
number of trichloroacetimidates have been used success-
fully to couple alkyl groups with the macrocycle.1b In this
approach, the active N-H in imidazole 11 is first
protected and the ester is reduced to the alcohol. The
alcohol is then activated by conversion to the trichloro-
acetimidate 12 (Scheme 3) to be coupled to the macro-
cycle. In analysis of the requirements for the imidazole
protecting group for the coupling reaction, it was believed
that an electron-donating group on the nitrogen might
facilitate the trichloroacetimidate cleavage to form the
carbonium ion intermediate 13A,B shown in Scheme 3.
azole nitrogen of 5 in the presence of the alcohol, this
compound is not very useful. When chloroacetaldehyde
was used, only the hydrated oxazole 7 was isolated. With
glyoxalic acid 8, only the decarboxylated imidazole 9 was
isolated in 61% yield. But when the hemiacetal of methyl
glyoxalate 10 was used, the desired aryl imidazole ester
11 was isolated in moderate yield. The major side product
was the imidazole 11A from the condensation of two
molecules of the aryl glyoxal with ammonia; thus an
excess amount of the methyl glyoxalate 10 had to be used.
Fortunately, compound 10 is commercially available at
low cost. After considerable optimization, it was found
that the best yield for 11 was obtained when the reaction
was carried out at neutral pH near 0 °C. As the solvent,
the acetonitrile-water mixture was found to give the
highest yield and reproducibility. The reaction was
carried out by adding a solution of the aryl glyoxal 1 and
Preliminary results in the coupling reaction of cyclohex-
- 6a
anol with SEM (Me3SiCH2CH2OCH2
)
and p-toluene-
sulfonyl-protected imidazole (12A where PG ) SEM or
p-toluenesulfonyl) confirmed this hypothesis; that is, the
SEM-protected side chain gave a better coupling reaction
yield than the tosyl-protected side chain. After some
(5) (a) Nakajima, N.; Horita, K.; Abe, R.; Yonemitsu, O. Tetrahedron
Lett. 1988, 33, 4139. (b) Schmidt, U.; Respondek, M.; Liebernecht, A.;
Werner, J .; Fisher, P. Synthesis 1989, 256.

Immunosuppressant L-733,725
J . Org. Chem., Vol. 64, No. 6, 1999 1861
Sch em e 4
Sch em e 5
preliminary study and taking into consideration the ease
of removal, the 2-tetrahydrofuranyl (THF) and 2-tetrahy-
dropyranyl (THP) groups were the final choices.6d,e Thus,
as shown in Scheme 4, the THF-protected imidazole 14
was prepared from dihydrofuran and the imidazole 11
with a catalytic amount of acid. Similarly, THP-protected
imidazole 17 was prepared from 11 and dihydropyran.
The ester groups in both compounds were then reduced
to the alcohols 15 and 18 in high yields with LiBH4.
The ease of removal of the THF and THP groups from
the imidazole alcohols 15 and 18 was investigated as a
model for the similarly protected L-733,725, i.e., the
expected coupling product. It is well established that
hydrolytic removal of this type of protecting group
depends on the electronic effects of the substituents on
the imidazole.6b The hydroxy groups on 15 and 18 should
be good models for the ether group in the expected
coupling product (protected L-733,725). At 50 °C with
excess tartaric acid in methanol, the half-life for the
removal of THF from 15 was 9 h, whereas the half-life
for removal of THP from 18 under the same conditions
was 54 h. Therefore, THF seemed to be the most suitable
protecting group. Thus the alcohol 15 was converted to
the trichloroacetimidate ether 16 with a catalytic amount
of DBU in 87% overall yield from the ester 11, as shown
in Scheme 4.5b Under optimized conditions, imidazole
ester 11 was converted to 14 and then reduced to 15 in
one pot, and crude 15 was used without purification. The
trichloroacetimidate compound 16 was isolated as an
easy to handle and stable white crystalline solid.
desired coupling product 21 was isolated in low yields
under a variety of conditions with several palladium and
nickel complexes. It was later found that the lithiation
was sluggish and never went to completion on the basis
of deuterium oxide trapping results. Thus when 20 was
treated with 2.2 equiv of n-butyllithium followed by D2O,
only 56% deuterium incorporation at the 5-position of 20
was observed.
Mod el Cou p lin g Rea ction . With the activated side
chain 16 in hand we studied its coupling reaction with
cyclohexanol as a model for ascomycin. In the literature,
the trichloroacetimidate-mediated coupling has been
developed for making benzyl or p-methoxybenzyl ethers
from alcohols. The most commonly used solvent is meth-
ylene chloride/cyclohexane, although diethyl ether has
also been used in some cases. Trifluromethanesulfonic
acid, boron trifluoride, methanesulfonic acid, and several
other acids have all been used as the catalyst.5
The alternate, aryl-aryl coupling route to the side
chain was also investigated briefly, as shown in Scheme
5. Thus imidazole was protected with the THF group and
then lithiated at the 2-position followed by trapping with
DMF and reduction, to give the 2-hydroxymethyl imid-
azole 20. This compound was then treated with 2 equiv
of butyllithium, transmetalated with magnesium bro-
mide, and treated with 3,5-dimethoxyphenyl iodide8 and
a catalytic amount of palladium or nickel complexes.7 The
In our first attempt, we used 10/1 CH2Cl2-diethyl
ether as the solvent and 1 equiv BF3-Et2O as the
catalyst. With 2 equiv cyclohexanol, a mixture of products
was observed. After column chromatography, only 7% of
the desired product 22 was isolated (eq 1). Other fractions
(6) (a) Brown, R. S.; Manoharan, T. S. J . Org. Chem. 1988, 53, 1107.
(b) Brown, R. S. Ulan, J . G. J . Am. Chem. Soc. 1983, 105, 2382. (c)
Brown, R. S.; Curtis, N. J . J . Org. Chem. 1980, 45, 4038. (d) Wu, D. C.
J .; Cheer, C. J .; Panzica, R. P.; Abushannab, E. J . Org. Chem. 1982,
47, 2661. (e) Hamamichi, N.; Migasaka, T. Tetrahedron Lett. 1985, 26,
4743.
(1)
(7) (a) Bell, A. S.; Roberts, D. A.; Ruddock, K. S. Tetrahedron Lett.
1988, 29, 5013. (b) Sekaiya, A.; Ishikawa, N. Organomet. Chem. 1977,
125, 281. (c) Tamao, K.; Sumitani, K.; Zembayashi, M.; Fujioka, A.;
Kodama, S.; Nakajima, I.; Minato, A.; Kumada, M. Bull. Chem. Soc.
J pn. 1976, 49, 1958. (f) Kalinin, V. N. Synthesis 1992, 413.
(8) Mann, I. S.; Widdowson, D. A. Tetrahedron 1991, 47, 7981.
from the column appeared to be oligomers (by NMR) from
the imidazole, presumably by Friedel-Crafts reaction

1862 J . Org. Chem., Vol. 64, No. 6, 1999
Song et al.
Ta ble 1. HP LC Assa y Yield (%) of 22 fr om Cou p lin g Rea ction betw een Cycloh exa n ol a n d 16 in Differ en t Solven ts (Yield
Ba sed on 16)
entry
solvents
yield
entry
solvents
yield
1
2
3
4
5
6
7
8
CH2Cl2/ether (10/1)
CH2Cl2/THF (10/1)
THF
MeCN/THF (10/1)
MeNO2/THF (10/1)
MeCN/i-PrCONMe2 (10/1)
MeCN/i-PrCONMe2 (1/1)
MeCN/t-ButCONMe2 (1/1)
24
32
18
49
51
58
65
68
9
10
11
12
13
14
MeCN/DMF (10/1)
MeCN/N-methylpyrrolidone (10/1)
MeCN/DMSO (10/1)
MeCN/THF/3-methyl-2-oxazolidinone (10/0.7/1)
MeCN/Me2NCONMe2 (10/1)
MeCN/DMPUa (10/1)
13
12
28
48
7
<1
a
DMPU: 1,3-Dimethyl-3,4,5,6-teatrahydro-2(1H)-pyrimidinone.
Sch em e 6
with the electron-rich dimethoxyphenyl ring. Mechanisti-
cally, the reaction presumably goes through a carbonium
ion intermediate, which can be trapped either by the
alcohol or by the dimethoxyphenyl ring (Scheme 3).
Apparently, in this case the selectivity is very low toward
ether formation.
Our initial strategy to enhance the selectivity was to
stabilize the carbonium ion intermediate by changing to
more polar and basic solvents.9 It is well established that
amides are more basic than ethers, which in turn are
more basic than nitriles and nitroalkanes.10 Obviously,
it would be necessary to find the solvent that has the
most appropriate basicity for this reaction. We started
with an extremely strong and nonnucleophilic acid,
HBF4-Et2O, which we believed would be the best for the
reaction, although it had not been used for this kind of
transformation before. The coupling reactions were done
with 1.4 equiv cyclohexanol and 1.0 equiv HBF4 at -10
°C in 1 h. The yields of 22 from the coupling reaction in
different solvents are summarized in Table 1. Because
of the limited solubility of the side chain and ascomycin
in some of these solvents, mixed solvents were used in
some cases. Methylene chloride, acetonitrile, and nitro-
methane were tried with ethers, amides, carbamate,
ureas, and DMSO as the cosolvents.
As seen from Table 1, great variations (<1% to 68%)
were observed with different solvents. Within certain
limits, the yield increases with increased basicity and
polarity of the solvent. Thus the yield increases according
to the order CH2Cl2-diethyl ether (entry 1) < CH2Cl2-
THF (entry 2) < MeCN-THF (entry 4) < MeCN-
dimethylpivalamide (N,N-dimethyl pivalamide) (entry 8).
But when the solvent is too basic as with tetramethyl
urea or DMPU (entries 14, 15), the reaction did not give
any desired product. Apparently, amides have the ap-
propriate basicity for this reaction. Upon screening
different amides, it was found that enolizable amides
such as N-methylpyrrolidinone or the reactive amide
DMF also interfere with the reaction. Carbamate (entry
12) was not better than amide. DMSO has a pKa similar
to amides, but the reaction yield in it was low. The best
solvent found was N,N-dimethylpivalamide mixed with
acetonitrile, wherein the yield was 68%.
Different acid catalysts were also compared in the
coupling reaction, and it was found that weak acids BF3
etherate and methanesulfonic acid gave much lower
yields than stronger protic acid HBF4 or trifuromethane-
sulfonic acid. When the protecting group was SEM
instead of THF, the coupling reaction with cyclohexanol
in acetonitrile-THF solvent gave similar results. How-
ever, when the protecting group was toluenesulfonyl, no
desired coupling reaction product with cyclohexanol could
be detected. On the basis of NMR evidence, the Friedel-
Crafts reaction on the dimethoxyphenyl ring was the
dominant reaction. These results confirmed the hypoth-
esis that electron-donating protecting groups can better
facilitate the coupling reaction.
From all the available data, a mechanistic explanation
for the dramatic effects of the solvents is proposed as
shown in Scheme 6. In the reaction the proton from the
acid preferentially protonates the more basic imidazole
nitrogen. The proton can equilibrate to the imidate
nitrogen to some extent, which leads to the carbon oxygen
bond cleavage to generate the carbonium intermediate.
When a basic solvent is used as in the case of dimethyl-
pivalamide, the carbonium ion can form a stabilized
intermediate like 23, which will react in a very different
manner from a less stabilized carbonium ion.11 When
DMF is the cosolvent, the resulting intermediate 24 could
react with alcohol at the less hindered formamide carbon.
In the case of N-methylpyrrolidone as cosolvent, enoli-
zation of the amide in the stabilized carbonium ion
intermediate could lead to side products, thus lowering
the yield of the desired coupling reaction. When a very
strongly basic solvent like tetramethylurea is used, the
intermediate may be so stable that it resists any further
(9) (a) Bolton, J . L.; McClelland, R. A. Can. J . Chem. 1989, 67, 1139.
(b) Freedman, H. H. In Carbonium Ions; Olah, G. A., Schleyer, P. von
R., Eds.; Wiley-Interscience: New York, 1973; Vol. IV, pp 1501-1678.
(10) The known dieclectric constants (D) and pKa values of the
conjugate acids for some common solvents are taken from the following
references and listed in the format of solvent (D, pKa). H2O (78, -1.7),
CH2Cl2 (8.9, -), MeNO2 (35.9, -12), MeCN (37.5, -10), Me2CO (20.7,
-7), EtOAc (6, -6.5), Et2O (4.3, -3.6), THF (7.6, -2.1), PhCONMe2
(-, -1.6), tButCONMe2 (-, -0.43), MeCONMe2 (38, 0.96), DMF (-,
0), DMSO (-, 0), Me2NCONMe2 (23, 2.0). (a) Riddick, J . A. Organic
Solvents: Physical Properties and Methods of Purification; Wiley: New
York, 1986; p 536. (b) Arnett, E. M. Prog. Phys. Org. Chem.; 1963, 1,
324, (pKa for Me2NCONMe2 from Merck Index 1989; p 9160).
(11) J ulia, M.; Mestdagh, H. Tetrahedron 1983, 39, 433.

Immunosuppressant L-733,725
J . Org. Chem., Vol. 64, No. 6, 1999 1863
Sch em e 7
reaction with alcohol. O-Alkylated amide salts analogous
to the proposed intermediate 23 have been isolated and
characterized as reported in the literature.11
20s resin column in 85% recovery. The purity of L-733,725
at this stage was 95%, which is far below the normal
standard for drug manufacturing. Fortunately, the tar-
trate salt of L-733,725 was found to be stable and
crystalline, which is also a pharmocologically acceptable
salt. This crystallization by salt formation is a critical
step in large-scale manufacturing of the drug because it
allows rejection of many low-level impurities which are
otherwise impossible to control. Thus, the L-733,725 free
amine from the column was converted to the tartrate salt
in 94% yield. The purity of the final product was
consistently over 98%. The overall yield of L-733,725
tartrate salt is 45% from the side chain and 55% from
ascomycin (based on consumed ascomycin). This route
is much more scaleable than the existing route because
it requires only one resin column purification. The
number of reactions involving ascomycin is also mini-
mized to two, coupling and deprotection, thus reducing
the chance of producing similar, difficult to separate side
products.
Syn th esis of L-733,725. The best conditions for the
model coupling reaction were applied to the coupling of
the side chain 16 to the unprotected macrocycle ascomy-
cin. These conditions were further optimized relative to
stoichiometry, concentration, and reaction temperature.
Thus the reaction was carried out by adding 1.2 equiv of
trifluoromethanesulfonic acid to a solution of side chain
16 and 1.5 equiv of ascomycin in a 1/3 mixture of
acetonitrile-N,N-dimethyl pivalamide at -30 to 0 °C.
Hydrolytic removal of the THF group on the imidazole
was realized by addition of water to the reaction and
aging at 50 °C overnight. The HPLC assay yield of
L-733,725 at this point was 55% based on side chain and
66% based on consumed ascomycin. Most of the unre-
acted ascomycin can be recovered during the isolation
process. The major side product in the coupling reaction
was the Friedel-Crafts reaction product 26, which has
been isolated and identified. Ascomycin was moderately
overcharged to minimize this side product. Some self-
condensation of the side chain by Friedel-Crafts reaction
has also been observed. In an attempt to increase the
polarity of the reaction media, LiBF4 was added to the
reaction mixture (0.2 M LiBF4), but no yield increase was
observed. The alkylation reaction is apparently very
selective toward the C32 hydroxy group because no
alkylation of the less accessible C24 hydroxy and the C10
hydroxy (the lactol) of ascomycin was observed. This
unusually high chemoselectivity is the likely result of the
synergetic effects of the electron-donating protecting
group, the polar and moderately basic solvent, and the
strong acid catalyst. Earlier attempts to use the toluene-
sulfonyl-protected trichloroacetimidate imidazole in the
coupling reaction with BF3-Et2O as the acid catalyst did
not produce any detectable amount of the desired cou-
pling product in methylene chloride.
In conclusion, a practical, scaleable, convergent route
was developed for the synthesis of L-733,725. The prop-
erly activated and suitably protected side chain was made
in an efficient four-step process in 42% overall yield. The
unprotected macrocycle ascomycin was converted to
L-733,725 in a highly chemoselective alkylation reaction,
which is the result of the synergistic effects of the suitable
protecting group, the right solvent, and the proper acid
catalyst. Purification via resin column chromatography
followed by crystallization as the tartrate salt gave pure
L-733,725 suitable for clinical trials. This process has
been scaled up to make multi-kilogram quantities of the
final drug.
Exp er im en ta l Section
Reagents and solvents were used as received. All anhydrous
solvents were dried over molecular sieves as needed. Karl
Fisher (KF) titrations for water contents were carried out on
an Aquatest IV moisture analyzer. Melting points were
The crude L-733,725 free amine was isolated as an
amorphous solid from the product mixture via an HP-

1864 J . Org. Chem., Vol. 64, No. 6, 1999
Song et al.
obtained on a Thomas-Hoover melting point apparatus and
uncorrected. Thin-layer chromatography was carried on silica
gel 60 plates. Nuclear magnetic resonance (NMR) spectra were
obtained on a Bruker 250 MHz spectrometer. Chemical shifts
are reported as parts per million (ppm) downfield from
tetramethylsilane. Chemical ionization mass spectra (CIMS)
were observed with methane as the carrier gas. HPLC columns
are 4.6 × 250 mm size, and the packing particle size was 5
µm except as otherwise stated.
3′,5′-Dimethoxyacetophenone was custom-made by River-
side Organic Inc. N,N-Dimethylpivalamide was custom-made
by Chemo Dynamics Inc., or it can be made from 40%
dimethylamine in water and trimethyl acetyl chloride and
distilled under reduced pressure.
3′,5′-Dim eth oxyp h en ylglyoxa l Mon oh yd r a te (1). To a
round-bottom flask equipped with a nitrogen inlet, a gas outlet
to a bleach scrubber, an overhead stirrer, and a temperature
probe was charged DMSO (11.3L) and 3′,5′-dimethoxy-
acetophenone (1.50 kg, 8.32 mol). The solution was heated to
60 °C, and aqueous HBr (48%, 0.873 L, 7.72 mol) was added
slowly over 1.0 h via an addition funnel while maintaining
the reaction temperature between 60 and 68 °C. A nitrogen
sweep was employed to remove the dimethyl sulfide as it was
formed. Once the HBr addition was complete, the batch
temperature was maintained at 65 °C with external heating
until the reaction was complete (typically 6-8 h). The reaction
was quenched by adding the reaction mixture into water (20.9
L). The batch temperature was adjusted to 50 °C and extracted
with a mixture of toluene (1.67 L) and heptane (1.67 L). More
water (20.9 L) was added at 50 °C to the above aqueous layer,
and then the batch was cooled to 35 °C. It was seeded and
further cooled to 23 °C over 2 h and to 0 °C over another 2 h
and aged for 1 h to complete the crystallization. Filtration and
washing the wet cake with cold water (21.5 L) followed by air-
drying with suction for 24 h afforded 3.29 kg of the title
compound as an off-white wet solid, which was 41.7% pure by
weight (the rest is water, 77% yield). This wet cake can be
used directly in the next reaction. It can also be dried further
to constant weight as the monohydrate. Mp > 95 °C (dec).
HPLC: Zorbax SB-C18 (3 µm) 3.0 × 150 mm, at 30 °C, 30/70
to 80/20 acetonitrile-0.1% aqueous H3PO4 in 10 min, flow )
mL). This solution was extracted with 20 mL of ethyl acetate.
The aqueous layer was basified with 5 N sodium hydroxide to
pH ) 9 and extracted with ethyl acetate (20, 10 mL). The
combined ethyl acetate layer was dried over MgSO4. The
solution was concentrated to dryness to give the title compound
1
as an off-white solid, wt 0.22 g (61%). H NMR was identical
with 4(5)-phenyl imidazole authentic sample (Aldrich).
2-Ca r b om e t h oxy-4(5)-(3′5′-d im e t h oxyp h e n yl)im id -
a zole (11). To a reaction vessel equipped with a mechanical
stirrer and an internal cooling coil (glycol cooling) were added
water (1.0 L), ammonium acetate (1.15 kg, 15 mol), and
acetonitrile (10 L). The batch was cooled to 0 °C. In a separate
flask fitted with a stirrer was charged acetonitrile (19.5 L),
the glyoxal monohydrate 1 (2.56 kg wet cake, assay amount
1.06 kg, 5.0 mol), and methyl 2-hydroxy-2-methoxyacetate
(J anssen Chimica, 1.8 kg, 15 mol). All of the solid was
dissolved. Then methyl 2-hydroxy-2-methoxyacetate (10) (600
g) was added to the first reaction vessel. The glyoxal solution
from the flask was added with a continuous flow over 2.0 h,
while the reaction mixture was vigorously stirred and kept at
0 °C. Additional acetonitrile (500 mL) was used for rinse. The
reaction mixture was stirred for 30 min at 0-5 °C and 22 °C
for 1 h. It was mixed with isopropyl acetate (15 L) and washed
with 5% sodium bicarbonate (3 × 12 L, gas evolution!) and
brine (12 L). The top organic layer was concentrated in a
vacuum to ∼8.0 L (bath temp 0-45 °C, >25 in. vacuum). Some
of the imidazole product crystallized out. Then isopropyl
acetate (3 L) was added and the batch concentrated again to
8 L. This flushing was repeated three to five times until GC
showed that the solvent ratio was <8/100 (v/v) acetonitrile-
isopropyl acetate when the total volume was ∼11 L. The water
content of the supernatant should be less than 3800 µg/mL.
Then heptane (10 L) was added and the mixture stirred
overnight at 22 °C. The solid was collected by filtration and
washed with 1/1 heptane-isopropyl acetate (3 × 1 L) and
water (5 × 1 L). It was dried in 50 °C vacuum oven with
nitrogen sweep over 2 days to KF e 2000 µg/g. The title
compound was obtained as a tan solid, wt 850 g (63%). HPLC
98 wt % purity. TLC Rf ) 0.54 (ethyl acetate). HPLC: YMC
ODS-AM column; eluents A, MeCN; B, 0.1% phosphoric acid
in water; flow 1.5 mL/min. Time 0 A/B 30/70, 10 min 35/65,
20 min 80/20, 25 min 80/20, 26 min 30/70. UV detector at 220
nm. tR: 11 6.7 min, 11A 20.5 min, 1 4.7 min, glyoxal methanol
adduct 9.9 min, isopropyl acetate 7.3 min. Mp: 154-155 °C.
1H NMR (CDCl3): δ 7.49 (s, 1H), 6.93 (s, 2H), 6.44 (t, J ) 2.2
0.75 mL/min, UV detection at 220 nm, starting material tR
)
1
5.4 min, product 1 tR ) 2.1 min. H NMR (CD3CN): δ 7.19 (d,
J ) 2.3 Hz, 2H), 6.76 (t, J ) 2.3 Hz, 1H), 5.83 (t, J ) 8.4 Hz,
1H), 4.80 (d, J ) 8.4 Hz, 2H), 3.82 (s, 6H). 13C NMR (CD3CN):
196.4, 162.0, 136.2, 108.0, 106.8, 88.1, 56.3. CIMS MH+ - H2O
) 195.
Hz, 1H), 4.01 (s, 3H), 3.85 (s, 3H). 13C NMR (CDCl3
+
CH3COOH): 161.1, 159.9, 139.9, 136.9, 132.2, 122.2, 103.4,
Anal. Calcd for C10H12O5: C 56.60, H 5.70. Found: C 56.61,
H 5.83.
100.7, 44.4, 53.0. CIMS MH+ ) 263.
Anal. Calcd for C13H14N2O4: C 59.54, H 5.38, N 10.68.
Found: C 59.37, H 5.15, N 10.58.
2-C h lo r o m e t h y l-4-p h e n y l-5-h y d r o x y -2,5-d ih y d r o -
oxa zole (7). To a round-bottom flask charged with ammonium
acetate (0.77 g, 10 mmol), a 50% aqueous solution of 2-chloro-
acetaldehyde (0.78 g, 5 mmol), and acetic acid (10 mL) was
added dropwise a solution of phenylglyoxal monohydrate (0.76
g, 5.0 mmol) while the reaction mixture was stirred. The
reaction was stirred overnight, and half of the reaction mixture
was concentrated in a vacuum to ca. 5 mL. Solid precipitated
out, and it was collected by filtration. The solid was washed
with acetic acid (2 × 0.5 mL) and sucked dry, wt 0.21 g (40%
yield). Pure sample was obtained by recrystallization from
2-Ca r bom eth oxy-4(5)-p h en ylim id a zole (11B). The pro-
cedure used for preparation of this compound was the same
as for the last one, yield 58%. The compound was a white
crystalline solid, mp 185-188 °C. 1H NMR (CD3OD): δ 7.76-
7.90 (m, 2H), 7.60 (s, 1H), 7.2-7.4 (m, 3H), 3.95 (s, 3H). 13C
NMR (CD3OD): δ 159.9, 139.0, 129.80, 129.78, 126.5, 52.7. 13
C
NMR (CD3OD + CH3COOH): δ 159.9, 142.5, 139.0, 133.1,
129.8, 128.8, 126.5, 120.9, 52.8. CIMS ΜΗ+ ) 203.
1-(2′-Te t r a h yd r ofu r a n yl)-2-ca r b om e t h oxy-4-(3′′,5′′-
d im eth oxyp h en yl)im id a zole (14). To a 3 L round-bottom
flask charged with anhydrous THF (1 L) and 11 (50 g, 191
mmol) was added 2,3-dihydrofuran (26.6 g, 380 mmol). The
solution was heated to 50 °C, and then a solution of p-
toluenesulfonic acid monohydrate (1.0 g, 5 mmol) in 10 mL of
THF was added. The reaction mixture was stirred at 50 °C
for 60 min. After 2 h at 50 °C, the solution was cooled to room
temperature. This crude product solution was used directly
for the next step without purification. Otherwise, it can be
isolated after workup with ethyl acetate and sodium carbonate
solution and crystallization from a 1/1 ethyl acetate-hexane
mixture as a white solid in 82% yield. TLC Rf ) 0.58 (ethyl
acetate). HPLC: YMC ODS-AM column, eluents A ) MeCN,
B ) (95% 20 mM phosphate buffer pH ) 6.0, 5% MeCN), flow
1.0 mL/min, time 0 A/B 30/70, 10 min A/B 70/30, 20 min 70/
30, 21 min 30/70, UV detector at 220 nm. tR: 14 12.0 min, 11
1
THF-hexane as an amber-colored solid, mp 148-149 °C. H
NMR (DMSO-d6): δ 7.90 (d, J ) 6.7 Hz, 2H), 7.4-7.6 (m, 3H),
7.30 (d, J ) 8.1 Hz, 1H), 6.36 (dd, J ) 3.7, 8.0 Hz, 1H), 6.1 (m,
1H), 3.91 (d, J ) 3.2 Hz, 2H). 13C NMR (DMSO-d6): δ 168.4,
131.5, 129.7, 128.6, 128.4, 100.6, 99.8, 46.6. CIMS MH+ ) 212.
Anal. Calcd for C10H10ClNO2: C 56.75, H 4.76, N 6.62, Cl
16.75. Found: C 57.00, H 4.76, N 6.64, Cl 16.64.
4 (5)-P h en ylim idazole (9). To a round-bottom flask charged
with glyoxylic acid monohydrate (0.38 g, 5.0 mmol), ammonium
acetate (0.79 g, 10 mmol) in methanol (10 mL), and acetic acid
(5 mL) was added dropwise a solution of phenylglyoxal
monohydrate (0.38 g, 2.5 mmol) in methanol (12 mL) while
the solution in the flask was stirred. After 1.5 h, the reaction
mixture was concentrated in a vacuum to a minimum volume
of 3-5 mL and was mixed with 0.5 N hydrochloric acid (10

Immunosuppressant L-733,725
J . Org. Chem., Vol. 64, No. 6, 1999 1865
8.2 min, 2,3-dihydrofuran 7.6 min. Mp: 194-196 °C. 1H NMR
(CDCl3): δ 7.53 (s, 1H), 6.97 (d, J ) 2.3 Hz, 2H), 6.69 (dd, J )
2.2 Hz, 6.4 Hz) 6.4 (t, J ) 2.3 Hz, 1H), 4.30-4.40 (m, 1H),
4.05-4.10 (m, 1H), 3.97 (s, 3H), 3.84 (s, 3H), 2.50-2.65 (m,
1H), 1.9-2.2 (m, 3H). 13C NMR (CDCl3): δ 161.0, 159.6, 141.8,
135.0, 134.5, 117.6, 103.3, 100.1, 89.0, 70.36, 55.5, 52.5, 35.2,
23.5. CIMS MH+ ) 333.
transferred to a separation funnel and mixed with ethyl
acetate (30 mL) and saturated sodium bicarbonate (20 mL).
The two layers were separated, and the organic layer was
washed with water (10 mL). The organic layer was concen-
trated in a vacuum to a solid residue and redissolved in ethyl
acetate (20 mL). The solution was filtered through a pad of
silica gel. The filtrate was then concentrated in a vacuum to
a solid residue, which was triturated with ethyl acetate (4 mL)
and hexanes (4 mL). The product was collected by filtration
and was dried by sweeping air to a white solid, wt 0.24 g (35%),
mp 149-151 °C. A second crop was obtained from the mother
liquor by further crystallizing from ethyl acetate (1 mL) and
hexanes (3 mL), wt 0.19 g (28%, total 63% yield). TLC Rf )
0.65 (EtOAc). 1H NMR (CDCl3): δ 7.70 (s, 1H), 6.97 (d, J )
2.3 Hz, 2H), 6.38 (t, J ) 2.3 Hz, 1H), 6.2-6.3 (m, 1H), 4.1-4.3
(m, 1H), 3.96 (s, 3H), 3.82 (s, 6H), 3.7-3.8 (m, 1H), 1.95-2.2
(m, 2H), 1.6-1.9 (m, 4H). 13C NMR (CDCl3): δ 170.0, 159.5,
142.3, 134.9, 134.6, 117.6, 103.2, 100.4, 84.8, 68.9, 55.4, 52.5,
33.4, 24.5, 22.8. CIMS MH+ ) 347.
Anal. Calcd for C17H20N2O5: C 61.44, H 6.07, N 8.43.
Found: C 61.29, H 6.00, N 8.20.
1-(2′-Tet r a h yd r ofu r a n yl)-2-h yd r oxym et h yl-4-(3′′,5′′-
d im eth oxyp h en yl)im id a zole (15). The product solution of
14 in THF from last reaction was cooled to -5 °C under
nitrogen. To this reaction mixture (cloudy) was added lithium
borohydride (4.80 g, 220 mmol). Methanol (7.7 mL, 191 mmol)
in 10 mL of THF was added slowly over 10 min (caution:
hydrogen gas evolution!). This mixture was stirred at 10-15
°C for 2 h. Then an aqueous NH4Cl solution (200 g/L, 400 mL)
was added slowly (hydrogen gas!). The two layers were
separated, and the top organic layer was washed with 40 wt
% K2CO3 (300 L) solution and then brine (300 mL). This crude
product was used directly for the next step, or it can be isolated
by crystallization from a 1/1 THF-hexanes mixture as a white
solid in 68% for the two steps from the unprotected imidazole.
TLC Rf ) 0.16 (ethyl acetate). HPLC: same conditions as in
the last reaction, tR ) 8.0 min, mp 109-111 °C. 1H NMR
(CDCl3): δ 7.13 (s, 1H), 6.85 (d, J ) 2.3 Hz, 2H), 6.36 (t, J )
2.3 Hz, 1H), 6.08 (dd, J ) 2.8, 6.4 Hz, 1H), 5.05 (br.s, 1H),
4.77 (ABq, J ) 3.6 Hz, ∆ν ) 7.9 Hz, 2H), 4.05-4.15 (m, 1H),
3.90-4.00 (m, 1H), 3.83 (s, 6H), 2.2-2.4 (m, 1H), 2.0-2.2 (m,
3H). 13C NMR (CDCl3): δ 161.0, 147.4, 139.8, 135.6, 112.7,
Anal. Calcd for C18H22N2O5: C 62.42, H 6.40, N 8.09.
Found: C 62.45, H 6.41, N 7.80.
1-(2′-Tet r a h yd r op yr a n yl)-2-h yd r oxym et h yl-4-(3′′,5′′-
d im eth oxyp h en yl)im id a zole (18). It was prepared by re-
duction with lithium borohydride in the same procedure as
for 16 in 86% yield as a hygroscopic white solid, mp > 110 °C.
TLC Rf ) 0.28 (EtOAc). 1H NMR (CDCl3): δ 7.25 (s, 1H), 6.86
(d, J ) 2.3 Hz), 6.35 (t, J ) 2.3 Hz, 1H), 5.3-5.4 (m, 1H), 7.77
(ABq, J ) 13.6 Hz, ∆ν ) 16.0 Hz), 4.0-4.15 (m, 1H), 3.82 (s,
6H), 3.65-3.75 (m, 1H), 1.55-2.0 (m, 5H). 13C NMR (CDCl3):
δ 169.5, 147.4, 139.4, 135.5, 112.9, 102.7, 99.4, 83.1, 68.4, 56.7,
55.4, 31.4, 24.9, 22.8. CIMS MH+ ) 319.
102.82, 99.4, 85.6, 69.0, 56.8, 55.4, 32.6, 24.5. CIMS MH+
305.
)
Anal. Calcd for C17H22N2O4: C 64.13, H 6.97, N 8.8. Found:
C 63.84, H 6.89, N 8.64.
Anal. Calcd for C13H14N2O4: C 59.54, H 5.38, N 10.68.
Found: C 59.37, H 5.15, N 10.58.
2-Hyd r oxym eth yl-4 (5)-(3′′,5′′-d im eth oxyp h en yl)im id -
a zole (5). To a round-bottom flask charged with a solution of
11 (1.0 g, 3.8 mmol) in THF (15 mL) and methanol (0.28 g)
was added lithium borohydride solution in THF (2.9 mL 2.0
M). All of this was done under nitrogen. A slight exothermic
reaction and gas evolution were observed. More lithium
borohydride solution (2.9 mL) and methanol (0.18 g) were
added 1.5 h later. After one more hour, water (15 mL) and
ethyl acetate (15 mL) were added. The mixture was acidified
with 2 N HCl to pH ) 1, and the layers were separated. The
aqueous layer was basified to pH ) 8 with saturated sodium
bicarbonate and extracted with ethyl acetate (3 × 25 mL). The
ethyl acetate layers were combined and dried with sodium
sulfate. The residue after solvent removal was purified by flash
column chromatography (98/2 ethyl acetate-methanol) to gave
the title compound as an off-white solid, 0.32 g (36%), mp 193-
1-(2′-Tetr ah ydr ofu r an yl)-2-tr ich lor oacetim idoxym eth yl-
4-(3′′,5′′-d im eth oxyp h en yl)im id a zole (16). The solution of
15 from the last step was concentrated to ∼330 mL and flushed
with anhydrous ethyl acetate (KF < 200 µg/mL, 3 × 500 mL).
At the end, the residue volume was about 330 mL, and some
solid precipitated out. The KF of the supernant was 1000 µg/
mL. To this residue was charged ethyl acetate (1.4 L) to
dissolve most of the solid. To this mixture were added 25 g of
K2CO3 powder and trichloroacetonitrile (41 g, 287 mmol). The
mixture was stirred for 2 h. The solid was then removed by
filtration and the filtrate stirred with more K2CO3 powder (12
g) for 0.5 h followed by addition of DBU (0.92 g) in 20 mL of
ethyl acetate. The reaction mixture was aged at ambient
temperature until completion of reaction (2-4 h). Ethyl acetate
(1 L) was added to dissolve all of the product that precipitated
during the reaction. The mixture was filtered through Celite
(prewashed with ethyl acetate). The clear filtrate was concen-
trated under vacuum to a thick white slurry (530 g, ca. 550
mL). To this slurry was added heptane (750 mL), and the
slurry was stirred for 2 h. The solid was collected by filtration,
washed with heptane-ethyl acetate (3 × 50 mL 1.5/1) and
heptane (2 × 50 mL), and dried under sweeping nitrogen. The
title compound was obtained as a slightly tan, fluffy solid, 75
g (87% yield), mp 148-150 °C. HPLC showed 96.8 wt % pure,
>99 area %. TLC Rf ) 0.69 (ethyl acetate). HPLC: same
conditions as previous reaction tR ) 15.1 min. 1H NMR
(CDCl3): δ 8.54 (s, 1H), 7.32 (s, 1H), 6.94 (d, J ) 2.2 Hz, 2H),
6.38 (t, J ) 2.0 Hz, 1H), 6.10 (dd, J ) 2.6, 6.2 Hz, 1H)5.46 (s,
2H), 4.15-4.25 (m, 1H), 3.95-4.05 (m, 1H), 3.84 (s, 6H), 2.35-
2.55 (m, 1H), 2.05-2.30 (m, 3H). 13C NMR (CDCl3): δ 161.9,
161.0, 141.1, 135.8, 113.7, 102.8, 99.7, 90.9, 86.0, 69.2, 63.3,
55.4, 33.5, 24.5. CIMS MH+ ) 448.
1
193 °C. H NMR (CDCl3): δ 7.34 (s, 1H), 6.87 (d, J ) 2.1 Hz,
2H), 6.35 (t, J ) 2.1 Hz, 1H), 4.90 (s, 2H), 3.80 (s, 6H). 13C
NMR (CD3OD+CF3COOH): δ 163.1, 149.5, 135.0, 129.6, 115.9,
104.5, 102.3, 56.0, 55.8.
Anal. Calcd for C12H14N2O3: C 61.53, H 6.02, N 11.96.
Found: C 61.29, H 6.04, N 11.88.
Kin etics of Solvolytic Rem ova l of THP a n d THF
Gr ou p s fr om Im id a zole 15 a n d 18. In a typical procedure,
the THP- or THF-protected imidazole (0.03 mmol) and L-
tartaric acid (15 mg, 0.1 mmol) were dissolved in methanol
(1.0 mL) and water (0.1 mL). This solution was then stirred
at 50 °C, and the reaction was followed by HPLC to at least
50% reaction. The half-life time of the starting material was
by direct observation or was calculated by assuming the
reaction was first-order to the starting material. HPLC:
Zorbax phenyl column, eluents 40% MeCN, 60% (95% 20 mM
phosphate buffer pH ) 6.0 and 5% MeCN), flow 1.0 mL/min.
UV detector at 250 nm. tR: 5 4.9-5.2 min, 15 7.1 min, 18 9.5
min.
Anal. Calcd for C18H20Cl3N3O4: C 48.18, H 4.49, N 9.36 Cl
23.70. Found: C 48.09, H 4.37, N 9.43, Cl 24.10.
1-(2′-Te t r a h yd r op yr a n yl)-2-ca r b om e t h oxy-4-(3′′,5′′-
d im eth oxyp h en yl)im id a zole (17). To a round-bottom flask
charged with 11 (0.52 g, 2.0 mmol) and 3,4-dihydropyran (1.8
g, 22 mmol) in 10 mL of THF was added p-tolunesulfonic acid
monohydrate (20 mg). The solution was heated in a 60 °C oil
bath for 10 h, and TLC indicated complete reaction. After the
reaction mixture was cooled to room temperature, it was
1-(2′-Tetr a h yd r ofu r a n yl)-2-cycloh exylm eth yl-4-(3′′,5′′-
d im eth oxyp h en yl)im id a zole (22). Mod el Cou p lin g Rea c-
tion w ith Cycloh exa n ol. Typical procedure: To a dry two
neck round-bottom flask equipped with a nitrogen inlet, a
temperature probe, and a magnetic stirrer was added 16 (100
mg, 0.22 mmol, KF < 200 µg/g), cyclohexanol (31 mg, 0.31
mmol), anhydrous acetonitrile (3 mL), and N,N,2-trimethyl

1866 J . Org. Chem., Vol. 64, No. 6, 1999
Song et al.
acetamide (0.3 mL, dried over molecular sieves). This solution
was then cooled to -25 °C under nitrogen and HBF4Et2O
(Aldrich 85%, 41 mg, 36 µL, 0.22 mmol) was added via a
syringe. The reaction mixture was allowed to cool to -10 to 0
°C and stirred for 1.5 h. The yield of the reaction was
determined by taking aliquots and diluted with acetonitrile
and injecting into HPLC. The reaction mixture was worked
up with ethyl acetate and saturated sodium bicarbonate. Pure
coupling product was isolated after purification through silica
gel chromatography and eluted with 1/1 ethyl acetate and
hexanes. It was further purified by recrystallization from ethyl
acetate-hexanes. It is a white solid, mp 100-101 °C. TLC Rf
) 0.66 (EtOAc). HPLC: Zorbax phenyl column, eluents 70%
MeCN, 30% 0.02 M phosphate buffer pH ) 6, flow 1.0 mL/
min, UV detector at 250 nm. 22 tR) 7.2 min. 1H NMR
(CDCl3): δ 7.23 (s, 1H), 6.90 (d, J ) 2.3 Hz, 2H), 6.35 (t, J )
2.3 Hz, 1H), 6.18 (dd, J ) 6.9, 3.2 Hz, 1H), 4.68 (s, 1H), 4.15-
4.25 (m, 1H), 3.95-4.05 (m, 1H), 3.82 (s, 6H), 3.3-3.5 (m, 1H),
mixture was allowed to warm up to room temperature. The
red color faded gradually and a white precipitate was observed
as the temperature rose. Then 30 mL of methylene chloride
and Mg2SO4 was added to the mixture. It was stirred for 2 h
and filtered to remove all the solid. The filtrate was concen-
1
trated to an oily residue, wt 0.17 g. H NMR showed that the
peak at 6.93 ppm integrated as 0.44 H, i.e., 56% deuterium
incorporation. Some side products were also observed.
3,5-Dim eth oxy-1-iodoben zen e.8 It was made from 1-chloro-
3,5-dimethoxybenzene by first converting the chloride to the
Grignard reagent (in THF), then quenching the Grignard
reagent with I2. It was purified by recrystallization from
1
hexane. H NMR (CDCl3): δ 6.85 (d, J ) 2.3 Hz, 2H), 6.40 (t,
J ) 2.3 Hz, 1H), 3.76 (s, 6H).
1-(2′-Tet r a h yd r ofu r a n yl)-2-h yd r oxym et h yl-5-(3′′,5′′-
d im eth oxyp h en yl)im id a zole (21). To a 50 mL three-neck
round-bottom flask equipped with a temperature probe, a
magnetic stirrer, and a nitrogen inlet were added 20 (0.17 g,
1.0 mmol) and anhydrous THF (5 mL). This mixture was
cooled to -60 °C and degassed with nitrogen, and 1.5 M
n-butyllithium (1.46 mL, 2.2 mmol) in hexanes was added
dropwise. A brick red suspension developed. After 15 min at
-60 °C, magnesium bromide etherate (0.52 g, 2.0 mmol) was
added, and the mixture was warmed to room temperature. The
mixture became an amber-colored suspension. After it was at
room temperature for 15 min, palladium tetrakistriphenylphos-
phine (30 mg) was added. The mixture was degassed and filled
with nitrogen. Then 3,5-dimethoxy-1-iodobenzene (264 mg, 1.0
mmol) was added as a solution in 2 mL of anhydrous and
degassed THF. The mixture was stirred at 50 °C for 20 h. After
it was cooled to room temperature, the mixture was mixed with
ethyl acetate (20 mL), 0.l M EDTA disodium (10 mL), and
saturated ammonium chloride (20 mL). After the two layers
were separated, the aqueous layer was extracted with ethyl
acetate (20 mL) and methylene chloride (20 mL). The combined
organic layer was concentrated in a vacuum to remove all
solvents. The residue was mixed with methylene chloride (20
mL) and water (10 mL). After the two layers were separated,
the organic layer was concentrated to an oily residue, wt 350
2.35-2.5 (m, 1H), 1.1-2.3 (m, 13 H).13C NMR (CDCl3):
δ
161.0, 144.8, 140.3, 136.3, 113.0, 102.8, 99.4, 85.8, 77.2, 69.1,
62.6, 55.4, 33.4, 32.1, 32.0, 25.7, 24.5, 24.0. CIMS MH+ ) 387.
Anal. Calcd for C22H20N2O4: C 68.37, H 7.82, N 7.25.
Found: C 68.30, H 7.65, N 7.30.
1-(2′-Tet r a h yd r ofu r a n yl)im id a zole (19). To a flask
charged with imidazole (15 g, 0.22 mol) and imidazole hydro-
chloride (7.68 g, 0.074 mol) was added acetonitrile (100 mL)
and 2,3-dihydrofuran (42 g, 0.60 mol). This solution was heated
to 45 °C for 4 h and then cooled to room temperature. Most of
the solvent was then removed under vacuum, the residue was
then dissolved in water, and the pH was adjusted to 9 with
solid sodium carbonate. This aqueous solution was extracted
with methylene chloride (2 × 100 mL). After the methylene
chloride layer was dried and the solvent removed under
vacuum, a slightly tan oil was obtained, wt 19 g (47% yield).
It was used in the next step after drying azeotropically by
flushing under vacuum. 1H NMR (CDCl3): δ 7.60 (s, 1H), 7.07
(s, 1H), 7.02 (d, J ) 1.0 Hz, 1H), 6.98 (d, J ) 1.0 Hz, 1H), 5.89
(dd, 1H, J ) 3.0, 6.3 Hz), 3.93-4.14 (m, 2H), 2.0-2.5 (m, 4H).
13C NMR (CDCl3): δ 135.3, 129.9, 116.2, 86.6, 68.9, 33.0, 24.2.
CIMS MH+ ) 139.
1
mg. H NMR (CDCl3) indicated the molar ratio of the desired
1-(2′-Tet r a h yd r ofu r a n yl)-2-h yd r oxym et h ylim id a zole
(20). To a three-neck round-bottom flask charged with 19 (5.0
g, 36 mmol) in anhydrous THF (75 mL), cooled to -20 °C and
under nitrogen, was added 1.6 M n-BuLi (25 mL, 40 mmol)
solution in hexane. The reaction was aged at -20 °C for 20
min, and anhydrous DMF (2.9 g, 40 mmol) was added. The
reaction mixture was allowed to warm to room temperature
and stirred for 1 h. To this cloudy mixture was added ethanol
(75 mL), and the mixture was cooled to -15 °C. Then sodium
borohydride (1.37 g, 36 mmol) was added portionwise (exo-
thermic reaction). After the reaction mixture was stirred at
room temperature for 1 h, saturated ammonium chloride (50
mL) was added (some forming observed), and the mixture was
stirred overnight. Most of the solvents were removed under
vacuum to give a slurry residue, which was dissolved in a
minimum amount of water. This mixture was worked up with
methylene chloride and brine. From the methylene chloride
solution, 6.1 g of oil was obtained after solvent removal.
Crystallization from 5/1 ethyl acetate-hexane gave the title
compound as an off-white solid, 3.6 g (59% yield), mp 89-91
product to the starting iodide was 1.0/2.4 (30% conversion). A
small amount of the title compound was isolated from the
crude products through silica gel chromatography (eluted with
2% MeOH-EtOAc) of similar runs as a half solid. 1H NMR
(CDCl3): δ 6.84 (s, 1H), 6.49 (d, J ) 2.3 Hz, 2H), 6.44 (m, 1H),
6.2 (br. s, 1H), 4.74 (ABq, 2H), 4.05-4.10 (m, 1H), 3.7-1.9 (m,
1H), 3.76 (m 6H), 2.2-2.4 (m, 2H), 1.9-2.1 (m, 2H). 13C NMR
(CDCl3): δ 160.6, 148.7, 133.7, 131.9, 126.2, 108.0, 100.3, 86.5,
68.4, 60.4, 56.7, 55.4, 32.2, 25.2. CIMS MH+ ) 275.
L-733,725 Ta r tr a te: Cou p lin g Rea ction . To a 72 L
round-bottom flask was charged 16 (1.30 kg 2.90 mol), asco-
mycin (3.448 kg, 4.35 mol, 1.5 equiv, produced in house by
Merck, 98%), and N,N-dimethyl pivalamide (N,N-dimethyl
pivalamide, 19.5 L). The KF was 2400 µg/mL for the solution.
It was dried by adding acetonitrile (10 L, KF ) 100 µg/mL)
and removing it under vacuum at 15-25 °C/29 in. Hg vacuum).
The solvent flush with CH3CN was repeated twice (final KF
) 50 µg/mL). The reaction mixture was diluted with CH3CN
(6.5 L) and cooled to -33 °C under N2. Then trifluoromethane-
sulfonic acid (i.e., triflic acid, 522 g, 3.42 mol) was charged
into the batch. The reaction mixture was warmed to 0 °C over
3 h. Then water (6.5 L) was added, and the pH of the reaction
mixture was adjusted to 2-3 with triflic acid if it was higher.
The mixture was heated to 50 °C for 24 h. The mixture was
cooled to room temperature and mixed with ethyl acetate (13
L) and saturated sodium bicarbonate (6.5 L). The two layers
were separated, and the organic layer was washed with brine
(6.5 L). The combined aqueous layer was extracted with 6.5 L
of ethyl acetate. The combined organic layer was concentrated
in vacuo to a minimum volume and flushed with acetonitrile
(12 L, the residue had a KF ) 7% water). The residue was
flushed with 2-propanol (15 L and 5 L) to a KF ) 0.37% water.
The residue was then flushed with acetonitrile (10 L), diluted
with acetonitrile (40 L), and extracted with hexane (4 × 80
L). The desired L-733,725 stayed in the lower acetonitrile
1
°C. H NMR (CDCl3): δ 6.93 (d, J ) 1.3 Hz, 1H), 6.88 (d, J )
1.3 Hz, 1H), 6.13 (dd, J ) 3.0 Hz, 7.9 Hz, 1H), 5.4 (br, s, 1H),
4.69 (s, 2H), 4.09-4.18 (m, 1H), 3.95-4.05 (m, 1H), 3.36-2.55
(m, 1H), 2.05-2.3 (m, 3H). 13C NMR (CDCl3): δ 147.5, 126.7,
116.4, 85.5, 69.0, 55.9, 33.1, 24.5. CIMS MH+ ) 169.
Anal. Calcd for C8H12N2O2: C 57.13, H 7.19, N 16.66.
Found: C 56.88, H 7.09, N 16.65.
1-(2′-Tetr a h yd r ofu r a n yl)-2-h yd r oxym eth yl-5-d eu ter io-
im id a zole (20-d ). To a dry three-neck round-bottom flask
charged with 20 (168 mg, 1.0 mmol) in anhydrous THF (5 mL)
cooled to -65 to 78 °C was added dropwise 1.5 M butyllithium
(1.5 mL, 2.25 mmol) in hexanes under nitrogen. A intense brick
red color developed after 0.7 mL of butyllithium was added.
Some precipitate was also observed. This mixture was stirred
at -65 °C for 15 min, and then 60 µL of D2O was added. The

Immunosuppressant L-733,725
J . Org. Chem., Vol. 64, No. 6, 1999 1867
layer, and the dimethyl pivalamide was extracted into the
hexane layer. The acetonitrile layer was concentrated in vacuo
to a minimum volume as a thick brown oil, wt ) 12.8 kg. HPLC
analysis showed it contained 12.5% L-733,725 by weight. The
yield of L-733,725 was 1.60 kg (55% based on 16, 66% based
on ascomycin consumed). Other compounds in the mixture
include ascomycin (1.9 kg, 45% of the charge), the overalkyl-
ation product, i.e., 26 without the THF protecting groups (8.3%
relative to the L-733,725 peak at 215 nm), plus N,N-dimethyl
pivalamide. HPLC: YMC ODS-AM column at 50 °C, A MeCN,
B 0.1% H3PO4 in water, flow rate 1.2 mL/min. Time 0 A/B 30/
70, 10 min A/B 60/40, 40 min A/B 70/30, 50 min A/B 80/20, 51
min A/B 30/70. tR: THF-protected L-733,725 25 23.7 min,
ascomycin 26.4 min, overalkylation product 26 (THF protected)
35.1 min, L-733,725 17 min, deprotected 26 19.2 min.
Small amounts of the desired coupling reaction product 22
and the bisalkylation side product 23 were isolated by quench-
ing the coupling reaction mixture before water addition with
saturated sodium bicarbonate solution and separating the
organic products through silica gel column chromatography.
They were isolated in ∼90% purity and were partially char-
acterized. 25: 1H NMR (CDCl3): δ (peaks over 5.5 ppm) 7.22
(s, 1H), 6.88 (d, J ) 2.1 Hz, 2H), 6.32 (t, J ) 2.1 Hz, 1H), 8.26-
6.28 (m, 1H). Electron spray ionization MS M+/e ) 1078. 26:
1H NMR (CDCl3): δ (peaks over 5.5 ppm) 8.22-8.25 (m, 1H),
7.14 (s, 1H), 6.90 (m, 3H), 6.37 (d, J ) 2.5, 1H), 6.30 (m, 2H),
6.1-6.2 (m, 2H). Electron spray ionization MS M+/e ) 1364.
TLC (EtOAc) 25 Rf ) 0.43, 26 Rf ) 0.50.
HP -20S Resin P u r ifica tion . The HP-20S dry resin (32 L)
was swelled with acetone (40 L) for 1 h, then loaded onto a
chromatography column. The column was eluted with acetone
(72 L), acetonitrile (72 L), and 1/1 acetonitrile-water (72 L).
The column was allowed to age for 18 h. The final volume of
resin was 36 L. The crude batch of L-733,725 free base was
divided into two portions of 6.3 kg each. Each portion contained
787 g of L-733,725 by HPLC assay. One portion was dissolved
in 50/50 CH3CN-H2O (90 L), and this solution was loaded onto
the resin column at a rate of 2 bed volumes per hour. The
column eluent was collected in 36 L cuts (one bed volume).
Once loaded, the column was eluted with 50/50 CH3CN-H2O
up to fraction 14. The column was then eluted with 60/40
CH3CN-H2O from fractions 15-23. Fractions 4-7 contained
ascomycin, and fractions 14-23 contained L-733,725. Fractions
15-20 were combined, and solid NaCl (11 kg) was added. After
40 min of stirring, the layers were allowed to separate. The
upper organic layer (108 L) showed 756 g of L-733,725 by
HPLC assay. The organic layer was concentrated in vacuo to
afford an oil suspended in about 10 L of water. The batch was
extracted with ethyl acetate (10 L), and the organic was passed
through a pad of silica gel (1.5 kg). The aqueous was reex-
tracted with fresh ethyl acetate (2 × 10 L), and each extract
was filtered through the pad of silica gel. The filtrate was
concentrated in vacuo to afford a foam. The foam was dried
under vacuum (29 in. Hg/25 °C) for 18 h. The dry batch
weighed 716 g and was 95% pure by weight (86% recovery on
column). The column was washed by eluting it with acetone
(72 L) and acetonitrile (72 L) at a rate of 2-3 bed volumes per
hour. The column was reequilibrated with 50/50 acetonitrile-
water (72 L), allowed to age for 18 h, and used again.
L-733,725 Ta r tr a te Sa lt F or m a tion . In a 50 L flask
containing L-733,725 free base (1.45 kg at 80% purity, 1.16
kg, 1.15 mol) was added acetonitrile (7 L; KF ) 15 µg/mL),
ethyl acetate (6.5 L; KF ) 50 µg/mL), and water (140 mL) to
form a 1% aqueous solution. To the batch was added l-tartaric
((2R,3R)-(+)-tartaric acid, 215 g, 1.43 mol). The mixture in the
flask became cloudy within 10 min, and the tartaric acid
dissolved within 1 h. The batch was seeded with 4 g of
L-733,725 tartrate salt. The mixture was stirred at room
temperature, under N2 for 18 h. A sample of the mother liquors
showed 12.3 mg/mL of product. The thick white slurry was
collected by filtration and washed with pure ethyl acetate (2
× 1 L). It was dried first with sweeping air and then in a
vacuum oven on glass trays at 50 °C with a nitrogen sweep
for 8 h. This afforded 1.26 kg (94% recovery) of a white solid
which was 98.5 area % on HPLC. HPLC: YMC ODS-AM
column at 50 °C, eluent A MeCN, B 0.1% H3PO4 in water, time
0 A/B 50/50, 30 min A/B 80/20, flow ) 1.0 mL/min, UV
detection at 215 nm. L-733,725 exhibits three peaks on reverse
phase HPLC, a major peak of ∼95% and two minor equilibrium
peaks; major peak tR ) 17 min, minor peaks tR ) 16.2 min,
12.4 min, over alkylation product (deprotected 26) tR ) 19.4
min. The product area % includes the main peak and the
equilibrium species peaks. Mp ) 165.5 °C. [R]365 ) 607° (25
°C, 1.0% in MeOH). Two sets of peaks were observed on the
13C NMR of ascomycin and L-733,725. They are believed to
arise from two rotational isomers. 13C NMR (100.61 MHz,
acetone-d6, major rotational isomers): δ 211.9, 198.0, 173.3,
169.9, 166.2, 162.1, 147.7, 139.4, 138.9, 136.6, 133.1, 132.2,
124.7, 115.9, 103.4, 99.5, 98.1, 84.0, 82.5, 79.7, 76.2, 74.4, 73.6,
72.9, 70.3, 65.6, 57.4, 57.3, 57.2, 56.4, 55.59, 55.58, 49.8, 46.6,
41.1, 39.6, 36.9, 35.5, 35.4, 34.2, 33.4, 31.3, 30.7, 28.5, 26.9,
25.3, 25.2, 21.9, 20.2, 16.6, 16.0, 13.6, 11.9, 10.3. 1H NMR
(400.13 MHz, acetone-d6, selected data, major rotamer): δ 7.50
(s, 1H), 6.97 (d, J ) 2.3, 2H), 6.35 (t, J ) 2.3, 1H), 5.25 (d, J
) 4.8, 1H), 5.21 (br d, J ) 9.1, 1H), 4.95 (br d, J ) 10.3, 1H),
4.77 (s, 2H), 4.63 (br t, J ) 3.6, 1H), 4.54 (s, 2H), 4.34 (br d, J
) 13.1, 1H), 3,97 (m, 1H), 3.80 (s, 6H), 3.71 (dd, J ) 9.5, 1.2,
1H), 3.63 (m, 1H), 3.45 (s, 3H), 3.37 (s, 3H), 3.32 (s, 3H), 2.95
(td, J ) 13.1 3.2, 1H), 2.79 (dd, J ) 14.3, 5.6, 1H), 1.68 (d, J
) 1.2, 3H), 1.62 (d, J ) 1.2, 3H), 0.94 (d, J ) 6.3, 3H), 0.92 (d,
J ) 7.1, 3H), 0.90 (d, J ) 6.7, 3H), 0.82 (t, J ) 7.5, 3H).
Anal. Calcd for C59H87N3O20
: C 61.10, H 7.50, N 3.60.
Found: C 60.85, H 7.66, N 3.63. Electron spray ionization MS
M+/e ) 1008.
J O981805G