Asymmetric synthesis of some substituted-3-phenyl prolines

Article abstract of DOI:10.1016/S0014-827X(99)00052-X

The asymmetric synthesis of carboxyphenyl prolines was performed according to Schollkopf methodology, to prepare possible antagonists of the metabotropic glutamate receptor mGluR1.

Full text of DOI:10.1016/S0014-827X(99)00052-X

www.elsevier.nl/locate/farmac
Il Farmaco 54 (1999) 461–464
Asymmetric synthesis of some substituted-3-phenyl prolines
Fabrizio Micheli *, Romano Di Fabio, Carla Marchioro
Glaxo Wellcome S.p.A., Medicines Research Centre, Via Fleming 4, 37100 Verona, Italy
Received 29 January 1999; accepted 23 April 1999
Abstract
The asymmetric synthesis of carboxyphenyl prolines was performed according to Scho¨llkopf methodology, to prepare possible
antagonists of the metabotropic glutamate receptor mGluR1. © 1999 Elsevier Science S.A. All rights reserved.
Keywords: Metabotropic; Glutamate; Neurotransmitters; Carboxy-phenyl-glycines
During the last 15 years, glutamate has become
recognized as the main excitatory neurotransmitter
within the mammalian central nervous system (CNS),
playing different and fundamental roles in many differ-
ent aspects of the functioning of the CNS. The action
of glutamate occurs through both ionotropic receptors
(iGlu), that is through the activation of receptor-gated
cation channels (namely AMPA, Kainate and NMDA),
and through metabotropic receptors (mGlu), which are
seven transmembrane (7-TM) receptors coupled to G-
proteins [1]. The mGlu receptor family is actually quite
unusual, while retaining the characteristic structural
features of G-proteins coupled receptors—the seven
transmembrane domains—the members of this family
possess a huge N-terminal extracellular domain, in
which the glutamate binding site is thought to be
located [2,3]. The mGluRs family can be conveniently
divided into three groups on the basis of their sequence
homology and pharmacological behavior. Group I
(mGluR1 and R5) is positively coupled to phospholi-
pase C (PLC) while Group II (mGluR2 and R3) and
Group III (mGluR4,R6,R7 and R8) are negatively
coupled to adenilate cyclase (AC) [4]. Although site-di-
rected mutagenesis studies and molecular modeling
techniques shed some light on the theories concerning
the glutamate recognition domain on mGluRs recep-
tors, still relatively little is known about the interaction
of the binding sites with their ligands known in litera-
ture [3].
The search for potential novel agonists/antagonists at
the mGlu receptors was mainly towed by the synthesis
of a large number of phenylglycine derivatives which
showed some signs of potential interest even if their
receptor selectivity (among the different groups) is
sometimes not very high [5]. Some of the products
acting at the different receptors are depicted in Fig. 1.
As far as Group I receptors are concerned, the synthe-
sis of a potential antagonist could reveal interesting
therapeutic potentials for these products, literature
studies show that these molecules could be useful in
tackling different pathologies such as strokes, neurode-
generation, Parkinson’s disease, and pain [6].
Many academic researchers and pharmaceutical com-
panies are now involved in mGluR1 selective antago-
nists synthesis as shown by the large number of patents
and publications. Recently, Eli Lilly tried to mimic
CPGs replacing the phenolic component of 4C,3H-
CPG (3) with an indole ring [7], but unfortunately none
of the two derivatives proposed (6 and 7, depicted in
Fig. 2) were active as a mGluR1 antagonist up to a
concentration of 100 mM.
Fig. 1. Some of the known antagonist/agonist for the mGluRs family:
4-CPG (1), MCPG (2), 3H-4-CPG (3), 3,5-dihydroxy CPG (4) and
4-PPG (5).
* Corresponding author. Fax: +39-045-921 8196.
E-mail address: fm20244@glaxowellcome.co.uk (F. Micheli)
0014-827X/99/$ - see front matter © 1999 Elsevier Science S.A. All rights reserved.
PII: S0014-827X(99)00052-X

462
F. Micheli et al. / Il Farmaco 54 (1999) 461–464
Fig. 2. Eli Lilly indolo CPG analogs (6, 7), AIDA (8) and UPF 596
(9).
Fig. 4. Minimized conformers of known mGluR1 antagonists ((S)-4-
CPG, AIDA, UPF 596).
to obtain molecules active at the mGluR1 site. As a
tool to test this hypothesis we synthesized (S)-(3-car-
boxyphenyl)-3-proline (10, Fig. 3). This derivative and
its corresponding enantiomer (11, Fig. 3) are depicted
in Fig. 3.
Fig. 3. (S)-(3-Carboxyphenyl)-3-proline (10) and (R)-(3-carboxy-
phenyl)-3-proline (11).
A huge amount of work to modify and replace the
skeleton of the CPGs was done by Professor Pelliccia-
ri’s group at the University of Perugia. In the derivative
AIDA (8, Fig. 2), the aminoacidic moiety was cyclized
with the phenyl ring giving rise to a very selective
mGluR1 antagonist [8]. Moreover, the same research
group showed that the phenyl ring was not fundamen-
tal to the activity of the CPG synthesizing UPF 596 (9,
Fig. 2) in which a propellane ring replaced the aromatic
moiety [9]. As reported by these authors, the use of
molecular modeling would predict that the derivative 9
may interact with the described putative pharma-
cophoric points in the extracellular part of the mGluR1
receptor, even if it is considerably shorter than the
corresponding 4-CPG derivative [3]. The theory under-
lying the synthesis of these derivatives proposes that
not only a certain distance between the aminoacidic
group and the carboxylic moiety, but also a certain
degree of linearity is necessary between them in order
These molecules were submitted to a conformational
analysis within Sybyl in order to take account of their
rotable bonds [10] and the different conformers ob-
tained were minimized. The same work was performed
on the three known antagonists depicted in Fig. 4.
The conformer endowed with the lower energy of
(S)-(3-carboxyphenyl)-3-proline (10), has with the cor-
rect distance between the carboxy group on the aro-
matic moiety and the aminoacidic group to
superimpose with 4-CPG and AIDA. Obviously it can-
not be superimposed with the ‘shorter’ UPF 596 as
depicted in Fig. 5(D).
It can also be clearly observed that the product 10 is
not endowed with the supposed required linearity be-
tween the aminoacidic moiety and the carboxylic group
and that the two aromatic regions only partially
overlap.
The synthesis of this derivative was performed in
accordance with that described for the unsubstituted
Fig. 5. Superimposition of the lower energy conformer of derivative 10 (A) with (S)-4-CPG (B), AIDA (C) and UPF 596 (D).

F. Micheli et al. / Il Farmaco 54 (1999) 461–464
463
Scheme 1.
Scheme 2.
cinnamate in the literature by Scho¨llkopf et al. [11] and
is outlined in Scheme 1.
[12,13]. The final hydrolysis with hydrated lithium hy-
droxide in aqueous ethanol gave complete conversion
to the aminoacidic desired product 10 in high yield
after desalting.
The case reported here is the first in which an ethoxy-
carbonyl substituted cinnamate is used as a Michael ac-
ceptor with the Scho¨llkopf reagent. The complete
stereoselectivity of the attack can be explained according
to Scho¨llkopf’s notes on acrylates [11] and it is notewor-
thy that the presence of a second carbonyl group on the
aromatic moiety of the cinnamate probably does not
modify the proposed transition state [11].
In order to give chemical completion to this work, the
corresponding enantiomer 11 was also synthesized as re-
ported in Scheme 2. The enantiomeric purity for both the
enantiomers (10 and 11) were derived from the results of
chiral shift NMR techniques on both VI and VIb. The
tabulated spectra for compound 10 (11) and the experi-
mental conditions used for VI and VIb are reported in the
The 3%-ethoxycarbonylcynnamate (II), which was eas-
ily prepared via Heck reaction in high yield, was sub-
jected to attack by the commercially available (R)
i-propyl Scho¨llkopf reagent in dry THF under nitrogen
at −78°C to give intermediate III in 50% yield as
single product. The bis lactim derivative was hydro-
lyzed with 0.25 N HCl and the corresponding amino-
acid derivatives underwent an in situ cyclization to give
the corresponding lactam IV which was recovered in
60% yield after purification by flash chromatography.
Protection of the free carboxylic group with TMS–
CHN2 produced the protected derivative V in quantita-
tive yield. The chemoselective reduction of the g-lactam
with borane–dimethyl sulfide complex (6 equiv.) in
refluxing THF produced the intermediate VI in 70%
yield (no trace of the alcohol derived from a possible
reduction of the ethoxycarbonyl group was observed)

464
F. Micheli et al. / Il Farmaco 54 (1999) 461–464
footnote1. The possible racemization in the final step was
excluded once more with the help of NMR techniques (in
the case of racemization, syn products should have been
observed through Jcis).
References
[1] J.-P. Pin, R. Duvoisin, Neuropharmacology 34 (1995) 1.
[2] P.J. O’Hara, O.P. Sheppard, H. Tyøgersen, D. Veneziz, B.A.
Haldeman, V. McCrane, K.A. Houamed, C. Thomsen, T.L.
Gilbert, E.R. Mulvihill, Neuron 11 (1993) 41.
[3] G. Costantino, R. Pellicciari, J. Med. Chem. 39 (1996)
3998.
[4] P.J. Conn, J.-P. Pin, Annu. Rev. Pharmacol. Toxicol. 37 (1997)
205.
[5] J.C. Watkins, G.L. Collingridge, Trends Pharmacol. Sci. 15
(1994) 333.
[6] D.J. Madge, A.M. Batchelor, Annual Reports in Medicinal
Chemistry, Academic Press, New York, 1996, p. 31.
[7] B.P. Clark, J.R. Harris, Synth. Commun. 27 (1997) 4223.
[8] R. Pellicciari, R. Luneia, G. Costantino, M. Marinozzi, B.
Natalini, P. Jakobsen, A. Kanstrup, G. Lombardi, F. Moroni,
C. Thomsen, J. Med. Chem. 38 (1995) 3717.
[9] R. Pellicciari, R. Raimondo, M. Marinozzi, B. Natalini, G.
Costantino, C. Thomsen, J. Med. Chem. 39 (1996) 2874.
[10] Sybyl, Version 6.4, Tripos Associated, St. Louis.
[11] U. Scho¨llkopf, U. Pettig, U. Busse, E. Egert, M. Dyrbusch,
Synthesis (1986) 737.
The derivative 10 and its enantiomer 11 were tested in
CHO cells expressing rat-mGluR1 receptor, but unfortu-
nately none of them showed activity up to the concentra-
tion of 10 mM. This could be in accordance with the
above-described theory [3] and with the lack of linearity
between the two moieties involved within the recognition
of the putative pharmacophoric points, but the inactivity
could also be due to a degree of rotation around the
phenyl rotable bond greater than that expected by MM
calculation. To solve this problem for definite, further
work is in progress on more rigid derivatives.
Acknowledgements
The authors would like to thank the Chemistry and
the Pharmacology Departments of GlaxoWellcome
S.p.A.
[12] H.J. Monteiro, Synthesis (1974) 137.
[13] D.W. Young, C.M. Moody, Tetrahedron Lett. 35 (1994)
7277.
1 Compound 10 (11): 1H NMR (500 MHz, ppm, DMSO-d6): 13.0
(1H, bs), 8.90 (2H, bm), 7.94 (1H, m), 7.80 (1H, d), 7.62 (1H, d), 7.44
(1H, t), 3.67 (1H, d, J=7.5 Hz), 3.48 (1H, m) 3.3–3.1 (m, 2H), 2.3 (1H,
m), 1.9 (1H, m). The enantiomeric purity of VI and VIb was determined
at 500 MHz using chiral lanthanide shift reagents. Significant splitting
of the aromatic (Dl=8 Hz) and CH (3.6 ppm) (Dl=9 Hz) signals of
the enantiomers was obtained in CDCl3 solution, at 28°C, with Eu(-
dpm)3(tris[dipivaloylmethanate]europium(III)).
.