Complementary asymmetric routes to (R)-2-(7-hydroxy-2,3-dihydro-1 H-pyrrolo[1,2-a]indol-1-yl)acetate

Article abstract of DOI:10.1021/ol303070k

Two distinct and scalable enantioselective approaches to the tricyclic indole (R)-2-(7-hydroxy-2,3-dihydro-1H-pyrrolo[1,2-a]indol-1-yl)acetate, an important synthon for a preclinical S1P₁ receptor agonist, are reported. Route 1 employs a modified version of Smith's modular 2-substituted indole synthesis as the key transformation. Route 2 involves a highly enantioselective CuH-catalyzed 1,4-hydrosilylation as the stereodefining step. Both routes can be performed without chromatography to provide multigram quantities of the tricycle in ≥98% ee.

Full text of DOI:10.1021/ol303070k

ORGANIC
LETTERS
2012
Vol. 14, No. 24
6306–6309
Complementary Asymmetric Routes
to (R)‑2-(7-Hydroxy-2,3-dihydro‑1H‑
pyrrolo[1,2‑a]indol-1-yl)acetate
Thomas O. Schrader,* Benjamin R. Johnson,† Luis Lopez, Michelle Kasem,
Tawfik Gharbaoui,‡ Dipanjan Sengupta,‡ Daniel Buzard, Christine Basmadjian,§ and
Robert M. Jones
Department of Medicinal Chemistry, Arena Pharmaceuticals, 6154 Nancy Ridge Drive,
San Diego, California 92121, United States
tschrader@arenapharm.com
Received November 7, 2012
ABSTRACT
Two distinct and scalable enantioselective approaches to the tricyclic indole (R)-2-(7-hydroxy-2,3-dihydro-1H-pyrrolo[1,2-a]indol-1-yl)acetate, an
important synthon for a preclinical S1P1 receptor agonist, are reported. Route 1 employs a modified version of Smith’s modular 2-substituted
indole synthesis as the key transformation. Route 2 involves a highly enantioselective CuH-catalyzed 1,4-hydrosilylation as the stereodefining
step. Both routes can be performed without chromatography to provide multigram quantities of the tricycle in g98% ee.
Members of the S1P receptor subfamily of G-protein
coupled receptors represent highly attractive targets for
drug discovery due to the numerous biological functions
they elicit.1 In particular, the S1P1 receptor agonist FTY720
(1, Gilenya, Scheme 1) is currently used to treat relapsing
forms of multiple sclerosis (MS).2 Site-selective phosphor-
ylation of FTY720 (1) to its active metabolite FTY720-P (2)
results in S1P1 dependent sequestration of lymphocytes into
the lymphoid tissues, thereby limiting the concentration
of immune cells in circulation.3 This general mechanism of
immunomodulation may be useful for the treatment of
other autoimmune and neurodegenerative disorders.4
The therapeutic potential for S1P1 agonists has prompted
the development of a number of second generation S1P1
agonists with improved selectivity and safety profiles.5
An internal discovery program has identified a number of
novel tricyclic indoles (3) as potent and selective agonists
of the S1P1 receptor.6 Originally prepared as racemates,
separation of the individual enantiomers of 3 by chiral
(4) (a) Matsuura, M.; Imayoshi, T.; Chiba, K.; Okumoto, T. Inflamm.
Res. 2000, 49, 404–410. (b) Kitabayashi, H.; Isobe, M.; Watanabe, N.;
Suzuki, J.; Yazaki, Y.; Sekiguchi, M. J. Cardiovasc. Pharmacol. 2000, 35,
410–416. (c) Daniel, C.; Sartory, N.; Zahn, N.; Geisslinger, G.; Radeke,
H.; Stein, J. J. Immunol. 2007, 178, 2458–2468.
(5) (a) Bolli, M. H.; Lescop, C.; Nayler, O. Curr. Top. Med. Chem.
(Sharjah, United Arab Emirates) 2011, 11, 726–757. (b) Buzard, D. J.;
Thatte, J.; Lerner, M.; Edwards, J.; Jones, R. M. Expert Opin. Ther. Pat.
2008, 18, 1141–1159.
(6) (a) Buzard, D. J.; Han, S.; Lopez, L.; Kawasaki, A.; Moody, J.;
Thoresen, L.; Ullman, B.; Lehmann, J.; Calderon, I.; Zhu, X.; Gharbaoui,
T.; Sengupta, D.; Krishnan, A.; Gao, Y.; Edwards, J.; Barden, J.; Morgan,
M.; Usmani, K.; Chen, C.; Sadeque, A.; Thatte, J.; Solomon, M.; Fu, L.;
Whelan, K.; Liu, L.; Al-Shamma, H.; Gatlin, J.; Le, M.; Xing, C.; Espinola,
S.; Jones, R. M. Bioorg. Med. Chem. Lett. 2012, 22, 4404–4409. (b) Jones,
R. M.;Buzard, D. J.;Kawasaki, A. M.;Kim, S.;Thoresen, L.;Lehmann, J.;
Zhu, X. WO 2010027431, 2010.
† Present address: Takeda California, Inc., 10410 Science Center Drive,
San Diego, CA 92121, USA.
‡ Present address: Dart NeuroScience LLC, 7473 Lusk Boulevard,
San Diego, CA 92121, USA.
§ Present address: University of Strasbourg, 74 route du Rhin, BP 60024,
F-67401 Illkirch Cedex, France.
(1) Mutoh, T.; Rivera, R.; Chun, J. Br. J. Pharmacol. 2012, 165, 829–844.
(2) (a) Kiuchi, M.; Adachi, K.; Kohara, T.; Minoguchi, M.; Hanano,
T.; Aoki, Y.; Mishina, T.; Arita, M.; Nakao, N.; Ohtsuki, M.; Hoshino,
Y.; Teshima, K.; Chiba, K.; Sasaki, S.; Fujita, T. J. Med. Chem. 2000, 43,
2946–2961. (b) http://www.gilenya.com/info/multiple-sclerosis/multiple-
sclerosis-information.jsp.
(3) (a) Gonzalez-Cabrera, P. J.; Cahalan, S. M.; Ferguson, J.; Rosen,
H. Curr. Immunol. Rev. 2012, 8, 170–180. (b) Chiba, K.; Adachi, K.
Future Med. Chem. 2012, 4, 771–781.
r
10.1021/ol303070k
Published on Web 12/04/2012
2012 American Chemical Society

chromatography revealed that one enantiomer possessed
superior activity. To further investigate this chemical series
in preclinical studies, a scalable asymmetric synthesis of
chiral tricyclic indole precursor (R)-2-(7-hydroxy-2,3-dihy-
dro-1H-pyrrolo[1,2-a]indol-1-yl)acetate (4), or its enantio-
mer, was required. Because the absolute configurations of
the active enantiomers were unknown, a route incorporat-
ing the stereogenic carbon in a precursor of known config-
uration would serve as a stereochemical structure proof.
Two separate and complementary routes to tricycle 4 are
described herein.
dianion (6) with N-acyl oxazolidinone 510 gave the desired
N-Boc ketoaniline 8. TFA-mediated cyclization followed
by re-esterification of the deprotected tert-butyl carbox-
ylate with TMS-diazomethane gave indole 9 in an overall
21% yield (3 steps). The sequence was performed with little
optimization and, to our knowledge, represents the first
example of adding a dilithiated o-toluidine to an N-acyl
oxazolidinone. Further experimentation to improve yields
may significantly expand the scope of electrophiles em-
ployed in the modular indole synthesis protocol.
Scheme 2. Asymmetric Synthesis of Tricycle 4a
Scheme 1. FTY720 (1), FTY720-P (2), S1P1 Agonist Series 3,
and Chiral Precursor 4
The methodology developed in the laboratories of Amos
Smith in the mid-1980s for the construction of 2-substi-
tuted indoles served as the general approach to the target
molecule (4).7 Relevant to our work, in a later adaptation
of A. Smith’s work, Clark et al. performed an elegant
synthesis of the achiral parent tricyclic indole 7-methoxy-
2,3-dihydro-1H-pyrrolo[1,2-a]indole.8 The key two-step
transformation in Clark’s synthesis involved trapping the
dianion generated from N-(tert-butoxycarbonyl)-4-meth-
oxy-o-toluidine with the Weinreb amide of 4-chlorobuta-
noic acid, followed by TFA mediated cyclization to give
2-(3-chloropropyl)-5-methoxy-1H-indole. The full tricycle
was then formed by intramolecular nucleophilic displace-
ment of the tethered chloro substituent by the indole
nitrogen. Our first generation asymmetric synthesis of 4,
shown in Scheme 2, is an extension of these methods where
the electrophile employed in the key dilithiation reaction
is the readily available chiral N-acyl oxazolidinone 5.9
Treatment of A ring precursor 7 (1.2 equiv) with s-BuLi
(2.8 equiv) in THF at ꢀ40 °C followed by quench of the
The remaining steps involved closure of the C ring.
Removal of the benzyl ether of 9 by hydrogenolysis with
palladium hydroxide on carbon gave the free alcohol 10.
After mesylation of 10, deprotonation of the indole NꢀH
with sodium hydride in DMF effected ring closure. Finally,
deprotection of the phenolic silyl ether with TBAF gave
ester 4a in excellent yield (96%, 3 steps). The high enantio-
meric purity of 4a (90% ee) revealed that no epimerization
of the stereocenter had occurred.11 The robust nature of
this protocol highlights its relevance in the synthesis of
2-substituted indoles with a tertiary stereogenic center
adjacent to the indole 2-position.12 Further elaboration
of tricycle 4a to S1P1 agonists (3) allowed for assignment of
the absolute stereochemistry of these agonists (3).
(10) N-Acyl oxazolidinone 5 was prepared in 90% diastereomeric
excess by stereoselective alkylation of (S)-4-benzyl-3-(4-(benzyloxy)-
butanoyl)oxazolidin-2-one. See: (a) Evans, D. A.; Wu, L. D.; Wiener,
J. J. M.; Johnson, J. S.; Ripin, D. H. B.; Tedrow, J. S. J. Org. Chem. 1999,
64, 6411–6417. The synthesis of compound 5 was reported previously,
but no analytical data were given. See: (b) Hepperle, M. E.; Campbell,
D. A.; Winn, D. T.; Betancort, J. M. WO 2009102876, 2009.
(11) Another significant feature of N-acyl oxazolidinone 5 is that
epimerization at the R-stereogenic carbon in these systems is strongly
discouraged, suggesting that the R-stereochemistry at the asymmetric
carbon atom would remain unaffected by the strongly basic condi-
tions encountered during the dilithiation reaction. See: Evans, D. A.
Aldrichimica Acta 1982, 15, 23–32.
(12) The Smith indole synthesis protocol has been successfully ap-
plied to formal total syntheses of (þ)-cinchonamine and (þ)-epi-cinch-
onamine in which tertiary asymmetric carbon atoms adjacent to the
indole 2-position are also located at the C2 position of bridged bicyclic
quinuclidine rings. In both cases the indole formation proceeded by
intramolecular heteroatom Peterson olefination. No epimerization was
observed. For details, see ref 7b.
(7) In Smith’s modular indole synthesis protocol a dianion generated
from an N-trimethylsilyl-o-toluidine is trapped with an ester or lactone
and the resulting N-lithio ketoaniline undergoes intramolecular heteroa-
tom Peterson olefination to give a 2-substituted indole. See: (a) Smith,
A. B., III; Visnick, M. Tetrahedron Lett. 1985, 26, 3757–3760. (b) Smith,
A. B., III; Visnick, M.; Haseltine, J. N.; Sprengeler, P. A. Tetrahedron
1986, 42, 2957–2969. In some cases the heteroatom Peterson olefination
does not proceed and the indole is produced by acid-promoted dehydra-
tion of the protonated ketoaniline intermediate. For an example, see:
€
(c) Smith, A. B., III; Davulcu, A. H.; Cho, Y. S.; Ohmoto, K.; Kurti, L.;
Ishiyama, H. J. Org. Chem. 2007, 72, 4596–4610.
(8) (a) Clark, R. D.; Muchowski, J. M.; Fisher, L. E.; Flippin, L. A.;
Repke, D. B.; Souchet, M. Synthesis 1991, 10, 871–878. For a similar
example, see also: (b) Peters, R.; Waldmeier, P.; Joncour, A. Org.
Process Res. Dev. 2005, 9, 508–512.
(9) It was reasoned that selective nucleophilic addition of a carbanion
to the exocyclic carbonyl adjacent to the oxazolidinone in 5 would be
favored both sterically and stereoelectronically over addition to the
tert-butyl ester.
Org. Lett., Vol. 14, No. 24, 2012
6307

After some preliminary attempts to scale up the route, a
number of modifications became necessary. It was found
that TFA-mediated cyclization of N-Boc ketoaniline 8 to
the indole was less successful on multigram scale.13 Treat-
ment with TMSI proved to be an acceptable means of
cyclization to the indole. However, the highly acidic nature
of this reagent caused deprotection of the TBS-protected
phenol. Exchange for the more stable TIPS-protected
phenol was warranted. Additionally, in order to avoid
tedious purifications to remove the chiral auxiliary, methyl
ester 1114 was examined as the electrophile in the dilithia-
tion reaction (Scheme 3). In initial experiments on small
scale (<1 g), nucleophiles generated by dilithiation of
either 7 or the TIPS protected variant 12 were quenched
with the chiral acyclic ester 11 (91% ee) at ꢀ78 °C to
give the desired N-Boc ketoaniline 8 or 14 respectively.15
Analysis of the reaction mixtures of 8 and 14 by 1H NMR
spectroscopy and LC/MS after aqueous workup indicated
selective addition of the benzyllithium anions (6 or 13) to
the desired methyl ester carbonyl of bis(ester) 11. When
performed on 99 g scale, N-Boc-o-toluidine 12 (1.6 equiv)
was treated with s-BuLi (3.7 equiv) at ꢀ40 °C and the
resulting dianion (13) was quenched with bis(ester) 11 at
ꢀ78 °C to deliver ketone 14. TMSI-promoted cyclization
and re-esterification of the hydrolyzed tert-butyl carbox-
ylate under mild conditions gave methyl ester 15 in 32%
overall yield from bis(ester) 11.16
The final steps to form the C ring were performed on
scale as described previously. The benzyl ether was
removed using 10% palladium on carbon under 55 psi H2
to give the free alcohol 16 in near-quantitative yield. Activa-
tion of 16 as its mesylate, C ring closure by deprotonation
of the indole with sodium hydride, and deprotection of
the phenolic TIPS-ether with TBAF gave a mixture of the
final tricycle (4a) and the free acid (17).17 Treatment of
the mixture of acid (17) and methyl ester (4a) with TMS-
diazomethane converted any free acid (17) to the ester (4a).
Compound 4a was obtained in 63% yield (4 steps from 16)
and 92% ee. Recrystallization of the product (4a) further
improved the enantiopurity (54% yield, 98% ee). In total,
over 12 g of 4a were prepared from bis(ester) 11 in a single
batch without the use of chromatography. This sequence
demonstrated the successful application of an acyclic ester
containing a tertiary asymmetric carbon atom at the
R-position as the electrophilic component in the modular
indole synthesis protocol. This highly useful intermediate
Scheme 3. Scale-up Synthesis of Tricycle 4a
(4a) was used to prepare a number of important S1P1
agonists (3) for examination in preclinical studies.
Concurrent with the development of the dilithiation
route, a second generation asymmetric route to compound
4 was being investigated. While the dilithiation route was
able to serve as a stereochemical structure proof and provide
material for preclinical testing of a number of advanced
S1P1 agonists (3), an inexpensive and more efficient route to
our most advanced drug candidate was desired.
The second approach to tricycle 4 was based on the
original medicinal chemistry route.6 An optimized route
to a key intermediate, the β,β-disubstituted enoate 18, as
a pure E-isomer is shown in Scheme 4.18 Starting from
readily available hydrazine 19, ethyl 5-(benzyloxy)-1H-
indole-2-carboxylate (20) was prepared via the classic
Fischer-indole approach.19 1,4-Addition/Dieckmann con-
densation of 20 with ethyl acrylate and KOt-Bu in THF at
reflux followed by decarboxylation in acetic acid/water
gave pure ketone 21 in 60% yield after precipitation from
the cooled reaction mixture. Olefination of the ketone (21)
with triethylphosphonoacetate (22) and KOt-Bu proceeded
with excellent E/Z selectivity (E/Z >35:1) to give the
R,β-unsaturated ester 18 exclusively as the E-isomer. Each
reaction in this optimized linear sequence was performed
on >0.30 mol scale without the use of chromatography.
With a convenient route to pure olefin isomer 18 estab-
lished, efforts were focused on identifying an asymmetric
reduction of 18. Because of the highly unique nature of
the substrate (18), there was little precedent to achieve
this transformation.20 Asymmetric hydrogenations of β,β-
disubstituted enoates lacking an adequate coordinating
group involve high catalyst loadings and hydrogen pres-
sures in excess of 500 psig and often proceed with low
(13) Similar results were observed by the F. Hoffman-La Roche
group during scale up of their 5-HT2c agonist precursor. See ref 8b.
(14) Bis(ester) 11 was prepared in two steps from N-acyl oxazolidi-
none 5 by peroxolysis followed by methylation. For removal of the chiral
auxiliary by lithium hydroxide/hydrogen peroxide, see: Evans, D. A.;
Britton, T. C.; Ellman, J. A. Tetrahedron Lett. 1987, 28, 6141–6144. Also
see Supporting Information.
(15) It is noteworthy Clark et al. reported in ref that attempts to
effect the reactions of dilithiated N-Boc-o-toluidines with esters were
unsuccessful. For an example of addition of a poly-lithiated N-(tert-
butoxycarbonyl)-o-toluidine to a lactone with a quaternary R-stereogenic
center, see ref 7c.
(18) The medicinal chemistry route had produced tert-butyl 2-(7-
hydroxy-2,3-dihydro-1H-pyrrolo[1,2-a]indol-1-ylidene)acetate as an
inseparable mixture of olefin isomers. See ref 6b.
(19) Ethyl 5-(benzyloxy)-1H-indole-2-carboxylate (20) could be pur-
chased from Amfinecom, Inc. for $6,800/kg. The hydrazine precursor
(19) was purchased from Astatech, Inc. for only $700/kg.
(16) In some cases, on 1ꢀ2 g scale, the TMSI conditions caused
deprotection of the TIPS ether.
(17) Some ester hydrolysis occurred during workup of the C ring
cyclization.
6308
Org. Lett., Vol. 14, No. 24, 2012

Scheme 4. Large-Scale Preparation of R,β-Unsaturated Ester 18
Scheme 5. Large Scale Asymmetric 1,4-Hydrosilylation
THF was superior to PhMe in solubilizing the enoate (18).
Ultimately, the CuH catalyzed reduction proved to be quite
scalable. When performed on a 90 g scale with 0.5 mol %
of (R,S)-PPF-P(t-Bu)2 (23) and 0.5 mol % Cu(OAc)2 with
THF as solvent, the reaction gave 4b in overall 79% yield
after benzyl group removal. The sequence was also per-
formed on a 250 g scale with a two-step yield of 67%.24
Both batches produced the tricycle (4b) in >98% ee after
recrystallization. The remarkable enantioselectivity, reac-
tion efficiency, and scalability of the asymmetric 1,4-hydro-
silylation of enoate 18 clearly demonstrates the utility of this
technology in preclinical drug development.
In conclusion, two complementary and scalable routes
to the S1P1 tricyclic agonist precursor, (R)-2-(7-hydroxy-
2,3-dihydro-1H-pyrrolo[1,2-a]indol-1-yl)acetate (4), have
been developed. The first generation dilithiation route
served as a stereochemical structure proof and provided
material for characterizing a number of promising S1P1
agonist drug candidates (3). The second route showcased
a highly enantioselective CuH-catalyzed asymmetric 1,4-
hydrosilylation to generate the stereogenic center. In total,
the optimized route produced in excess of 178 g of tricycle
4b in >98% ee without the use of chromatography. The
material obtained enabled a number of preclinical studies
for a highly advanced drug candidate.
enantioselectivities.21 After some preliminary screening of
conditions to reduce 18, we decided to further examine the
CuH-catalyzed asymmetric 1,4-hydrosilylation (Scheme 5)
as an alternative to hydrogenation.22 In this procedure,
a CuH complex ligated with a chiral phosphine is the active
catalyst and a silane serves as the stoichiometric hydride
donor. High substrate/ligand ratios and the use of envir-
onmentally benign and inexpensive hydride sources such
as tetramethyldisiloxane (TMDS) and polymethylhydro-
siloxane (PHMS) make this an attractive procedure for scale
up. A screen of chiral phosphines, solvents, and reaction
temperatures was performed on small scale (100 mg of 18)
using PHMS as the hydride source and t-BuOH as an
additive.23 The ferrocenyl based JOSIPHOS ligand (R,S)-
PPF-P(t-Bu)2 (23) and (R)-DTBM-SEGPHOS gave super-
iorresults(95ꢀ99% ee) to the BINAP based bis-phosphines
(R)-BINAP and (R)-xylyl-BINAP (69ꢀ89% ee). The tem-
perature and solvent seemed to have little effect, although
(20) Subsequent to this work, the iridium-Walphos-catalyzed enan-
tioselective hydrogenation of a related tricyclic enoate was reported. See:
Tudge, M.; Savarin, C. G.; DiFelice, K.; Maligres, P.; Humphrey, G.;
Reamer, B.; Tellers, D. M.; Hughes, D. Org. Process Res. Dev. 2010, 14,
787–798.
(21) (a) Tang, W.; Zhang, X. Chem. Rev. 2003, 103, 3029–3069.
Chiral iridium N,P complexes have performed well in the high pre-
ssure hydrogenations of unfunctionalized olefins. For reviews, see:
(b) Woodmansee, D. H.; Pfaltz, A. Chem. Commun. 2011, 47, 7912–
7916. (c) Minnaard, A. J.; Feringa, B. L.; Lefort, L.; de Vries, J. G. Acc.
Chem. Res. 2007, 40, 1267–1277.
(22) (a) Lipshutz, B. H.; Servesko, J. M.; Taft, B. R. J. Am. Chem.
Soc. 2004, 126, 8352–8353. (b) Deng, J.; Hu, X.-P.; Huang, J.-D.; Yu,
S.-B.; Wang, D.-Y.; Duan, Z.-C.; Zheng, Z. J. Am. Chem. Soc. 2008, 73,
6022–6024. (c) Deutsch, C.; Krause, N.; Lipshutz, B. H. Chem. Rev.
2008, 108, 2916–2927.
Acknowledgment. We would like to thank Professor
Jason S. Kingsbury (Pomona College) for providing useful
assistance in preparing this manuscript. We also thank
Walter Gwathney (Arena Pharmaceuticals) for performing
high-resolution mass spectra.
Supporting Information Available. Characterization
data and full experimental details. This material is avail-
able free of charge via the Internet at http://pubs.acs.org.
(23) See Supporting Information.
(24) The 250 g scale reduction proceeded more slowly and required
the infusion of additional CuH (1.5 mol %) and ligand 23 (1.5 mol %) as
well as a higher temperature (15 °C) in order for the reaction to go to
completion.
The authors declare no competing financial interest.
Org. Lett., Vol. 14, No. 24, 2012
6309