Stereoselective synthesis of annular 9-cis-retinoids and binding characterization to the retinoid X receptor

Article abstract of DOI:10.1021/jo0257391

Analogues of 9-cis-retinoic acid incorporating an alicyclic ring between the C19 and C10 positions have been synthesized and evaluated as ligands for the RXRα nuclear receptor. The stereocontrolled synthesis of these configurationally constrained retinoids combines a Stille cross-coupling and the Wittig reaction as key bond-forming steps. The palladium-catalyzed cross-coupling reaction of the β-bromo-α,β-unsaturated aldehydes 5 to dienylstannane 6 is very fast at room temperature, and takes place with preservation of the dienylstannane geometry. A highly stereoselective Wittig reaction afforded the C7-C8 bond connecting the hydrophobic ring to the retinoid side chain. The binding affinities of these compounds for the receptor were determined, and the structural and energetic rationale behind the affinity profile of the cyclic 9-cis-retinoic acid derivatives for the RXRα nuclear receptor was characterized by using Molecular Mechanics protocols.

Full text of DOI:10.1021/jo0257391

Ster eoselective Syn th esis of An n u la r 9-cis-Retin oid s a n d Bin d in g
Ch a r a cter iza tion to th e Retin oid X Recep tor
M. Paz Otero, Alicia Torrado, Yolanda Pazos, Fredy Sussman,‡ and Angel R. de Lera*
Departamento de Quı´mica Orga´nica, Facultad de Ciencias, Universidade de Vigo, 36200 Vigo, Spain, and
Departamento de Quı´mica Orga´nica, Facultad de Qu´ımica, Universidade de Santiago de Compostela,
Santiago de Compostela 15706, Spain
qolera@uvigo.es
Received March 20, 2002
Analogues of 9-cis-retinoic acid incorporating an alicyclic ring between the C19 and C10 positions
have been synthesized and evaluated as ligands for the RXRR nuclear receptor. The stereocontrolled
synthesis of these configurationally constrained retinoids combines a Stille cross-coupling and the
Wittig reaction as key bond-forming steps. The palladium-catalyzed cross-coupling reaction of the
â-bromo-R,â-unsaturated aldehydes 5 to dienylstannane 6 is very fast at room temperature, and
takes place with preservation of the dienylstannane geometry. A highly stereoselective Wittig
reaction afforded the C7-C8 bond connecting the hydrophobic ring to the retinoid side chain. The
binding affinities of these compounds for the receptor were determined, and the structural and
energetic rationale behind the affinity profile of the cyclic 9-cis-retinoic acid derivatives for the
RXRR nuclear receptor was characterized by using Molecular Mechanics protocols.
In tr od u ction
by both trans-retinoic acid 1 and 9-cis-retinoic acid 2
(Scheme 1) whereas only the latter ligand activates the
RXRs (isotypes RXRR, RXRâ, and RXRγ). Moreover, the
RARs share a common mechanism of gene transcription
with other members of the nuclear receptor superfamily2
(vitamin D receptor (VDR), thyroid hormone receptor
(TR), peroxisome proliferator-activated receptors (PPARs),
and liver X receptor (LXR)) activated by small lipophilic
ligands (vitamin D3, thyroid hormone, fatty acids, and
sterols, respectively), namely the formation of het-
erodimers with the RXRs. For retinoid signaling, the
dimeric form of RXR/RAR binds to DNA sequences called
retinoic acid response elements (RAREs) located in the
promotor region of target genes. On the other hand, the
RXRs can also act autonomously as homodimers and bind
to 9-cis-retinoic acid response elements (RXREs), thus
defining two different mechanisms by which retinoids
influence gene transcription.2
Native retinoids, the compounds related to vitamin A
found in cells, are essential dietary substances that are
needed by mammals for reproduction, normal embryo-
genesis, cell growth, maintenance of controlled cell
proliferation and differentiation, vision, and integrity of
the immune system.1 The recent discovery and charac-
terization of the retinoid receptors helped to understand
the developmental and physiologic networks governed by
vitamin A and its analogues.2 It is now well-established
that, within cells, retinoids regulate gene transcription
acting through binding to the evolutionary distinct group
of nuclear receptors called the retinoic acid receptors
(RARs) and retinoid X receptors (RXRs),1 which are
ligand-dependent transcription regulatory factors. The
RARs (isotypes RARR, RARâ, and RARγ) are activated
The finding that RXR is a critical component of
heterodimer formation has stimulated interest in the
pathways of intracellular formation of its ligand 9-cis-
retinoic acid 2.3 Although 9-cis-retinoids might be gener-
ated from dietary 9-cis-â-carotene, it is conceivable that
9-cis-retinoic acid 2 could be produced by intracellular
isomerization (perhaps enzymatically catalyzed) of the
thermodynamically more stable trans-retinoic acid 1. It
has recently been demonstrated that in bovine liver
membranes (presumably mediated by thiol-containing
catalysts) trans-retinoic acid 1 affords an equilibrium
mixture containing stereoisomers with differing geom-
etries around one and two double bonds, including 9-cis-
* Address correspondence to this author at Universidade de Vigo.
‡ Universidade de Santiago de Compostela.
(1) For monographs, see: (a) Sherman, M. I., Ed. Retinoids and Cell
Differentiation; CRC Press: Boca Raton, FL, 1986. (b) Saurat, J . H.,
Ed. Retinoids: New Trends in Research and Therapy; S. Karger: Basel,
Switzerland, 1985. (c) Bernard, B. A.; Shroot, B., Eds. Retinoids: From
Molecular Biology to Therapeutics, Pharmacol. Skin; Karger: Basel,
Switzerland, 1993. (d) Livrea, M.; Vidali, G., Eds. Retinoids: From
Basic Science to Clinical Applications; Birkhau¨ser: Basel, Switzerland,
1994. (e) The Retinoids: Biology, Chemistry and Medicine, 2nd ed.;
Sporn, M. B., Roberts, A. B., Goodman, D. S., Eds.; Raven Press: New
York, 1993. For numbering of the retinoid structure, see: IUPAC-IUB
J oint Commission on Biochemical Nomenclature (J CBN). Eur. J .
Biochem. 1982, 129, 1.
(2) For reviews, see: (a) Evans, R. M. Science 1988, 240, 889. (b)
Mangelsdorf, D. A.; Evans, R. M. Cell 1995, 83, 841. (c) Katzenellen-
bogen, J . A.; Katzenellenbogen, B. S. Chem. Biol. 1996, 3, 529. (d)
Rosen, J .; Day, A.; J ones, T. K.; J ones, E. T. T.; Nadzan, A. M.; Stein,
R. B. J . Med. Chem. 1995, 38, 4855. (e) Weatherman, R. V.; Fletterick,
R. J .; Scanlan, T. S. Annu. Rev. Biochem. 1999, 68, 559. (f) Bourquet,
W.; Germain, P.; Gronemeyer, H. Trends Pharmacol. Sci. 2000, 21,
381.
(3) (a) Dawson, M. I.; Okamura, W. H., Eds. Chemistry and Biology
of Synthetic Retinoids; CRC Press: Boca Raton, FL, 1990. (b) Curley,
R. W.; Robarge, M. J . Curr. Med. Chem. 1996, 3, 325.
10.1021/jo0257391 CCC: $22.00 © 2002 American Chemical Society
Published on Web 07/09/2002
5876
J . Org. Chem. 2002, 67, 5876-5882

Stereoselective Synthesis of Annular 9-cis-Retinoids
SCHEME 1. Na tive Liga n d s of th e Retin oid Recep tor s a n d Discon n ection of th e 9-cis-Lock ed Retin oid s
retinoic acid 2.4 In addition, it is plausible that, after its
formation, the less stable 9-cis-retinoic acid 2 could be
partially converted in vivo into the trans-isomer 1, thus
inducing biological responses not easily ascribed to either
ligand. In this regard, 9-cis-retinoic acid analogues that
are unable to isomerize might, at the outset, shed light
into the relevance of in vivo isomerization of biologically
active retinoids, allowing at the same time the biological
effects attributable to 9-cis-retinoic acid 2 to be assessed.
With those considerations in mind, we initiated a
program aimed to design and synthesize configuration-
ally locked retinoic acid analogues,5,6 focusing on the
structure of the native RXR ligand 9-cis-retinoic acid 2.
The recently reported crystal structure of the ligand-
binding domain of RXR bound to its cognate ligand 9-cis-
retinoic acid 27 has revealed a sharp conformational
distortion of 2, with the polyene side chain past C9
positioned almost perpendicular to the hydrophobic ring.
The volume occupied by this ligand is only about 59% of
the binding pocket total volume, indicating that there is
additional space in the cavity for improving binding.
Unoccupied volume is available in two regions: near the
C19-methyl group close to Trp305 and Asn306, and in
the vicinity of the C18 methyl, close to Val349. We set
out to determine the effect on RXR binding affinities of
configurationally locked 9-cis-retinoic acid analogues
which incorporate an alicyclic ring near the C19 region.
The additional ring of annular 9-cis-retinoic acid ana-
logues (compounds 3) encompass the C19 and C10
positions, thus constraining the retinoid side-chain into
the 9-cis configuration, albeit an increase in ring size
might allow unsaturated rings with trans geometries to
be accommodated. Conformationally dependent RXR
binding affinities8 might then be assessed and molecular
modeling studies of the binding of the ring analogues to
the RXRR receptor could provide a rationale for the
observed results.
(4) Urbach, J .; Rando, R. R. Biochem. J . 1994, 299, 459.
(5) (a) Ikegami, S.; Iimori, T.; Kazama, M.; Ishii, H. J panese Kokai
Tokkyo Koho 1997, J P 09059207; Chem. Abstr. 1997, 126, 305659. (b)
Brooks, S. C.; Kazmer, S.; Levin, A. A.; Yen, A. Blood 1996, 87, 227.
(c) Minucci, S.; Saint-J eannet, J .-P.; Toyama, R.; Scita, G.; DeLuca, L.
M.; Taira, M.; Levin, A. A.; Ozato, K.; Dawid, I. B. Proc. Natl. Acad.
Sci. U.S.A. 1996, 93, 1803. The reported5a synthesis of 3a -c uses the
central building block 11, and the strategy combines also a single-
bond and a double-bond disconnection, using first a Suzuki coupling
Most of these configurationally locked 9-cis-retinoids
(3a -c) were first synthesized in view of their suspected
antithrombotic potential,5a and subsequently 3b has also
been shown to induce differentiation of HL-60 promie-
locytic leukemia cell lines.5b However, no RXR binding
selectivities and no comparative studies have been car-
ried out for these analogues, with the exception of 3b.5b
of boronate 10 and triflate 11, and then
Emmons condensation of 14 and 15.
a Horner-Wadsworth-
Resu lts a n d Discu ssion
Ch em istr y. Considering that the ring spanning the
C9 and C10 double bond is positioned at the central
region of the pentaene present in the desired targets 3,
we selected the disconnections shown in Scheme 1. The
combination of a single bond (C10-C11) and a double
bond (C7-C8) construction allows the convergent syn-
thesis of the retinoids by coupling known fragments 4,9
5,10 and 6.11
(8) Dawson, M. I.; J ong, L.; Hobbs, P. D.; Cameron, J . F.; Chao, W.;
Pfahl M.; Lee, Mi-Ock.; Shroot, B.; Pfahl, M. J . Med. Chem. 1995, 38,
3368.
(9) Zhang, H.; Lerro, K. A.; Takekuma, S.; Baek, D.; Moquin-Pattey,
C.; Boehm, M. F.; Nakanishi, K. J . Am. Chem. Soc. 1994, 116, 6823.
(10) Arnold, Z.; Holy, A. Collect. Czech. Chem. Commun. 1961, 25,
3051.
(11) (a) Yokokawa, F.; Hamada, Y.; Shioiri, T. Tetrahedron Lett.
1993, 34, 6559. (b) Betzer, J .-F.; Delagoge, F.; Muller, B.; Pancrazi,
A.; Prunet, J . J . Org. Chem. 1997, 62, 7768.
(6) For other examples of 9-cis-retinoids configurationally locked at
the same region with (hetero)aryl and cyclopropyl rings, see: (a) Apfel
C. M.; Kamber, M.; Klaus, M.; Mohr, P.; Davies, P. J . A. J . Biol. Chem.
1995, 270, 30765. (b) Vuligonda, V.; Garst, M. E.; Chandraratna, R.
A. S. Bioorg. Med. Chem. Lett. 1996, 6, 213. (c) Vuligonda, V.; Garst,
M. E.; Chandraratna, R. A. S. Bioorg. Med. Chem. Lett. 1999, 9, 589.
(7) Egea, P. F.; Mitschler, A.; Rochel, N.; Ruff, M.; Chambon, P.;
Moras, D. EMBO J . 2000, 19, 2592.
J . Org. Chem, Vol. 67, No. 17, 2002 5877

Otero et al.
SCHEME 2. Ster eocon tr olled Syn th esis of
The generation of the C7-C8 double bond in the
synthesis of retinoids poses no stereochemical concern
due to the high E-stereoselectivity of the common double
bond forming processes (Wittig, J ulia), a selectivity
enforced by the bulky trimethylcyclohexenyl ring.12 On
the other hand, the stereocontrolled C10-C11 single-
bond construction demands the use of organometallic
reactions. We have previously developed convergent
approaches to highly conjugated olefins with either side-
chain or ring modifications that afford geometrically
homogeneous retinoids (trans- or 9-cis), as required to
elicit biological responses.13 These approaches are based
on the palladium-catalyzed cross-coupling of alkenyl
partners, organometallics (boronic acids or stannanes),
and electrophiles (iodides, triflates). Both the Suzuki14
and the Stille15 coupling reactions display high levels of
regio- and stereoselectivity and the reaction conditions
are mild, which is especially well-suited for the prepara-
tion of unstable polyenes. For the case at hand, the Stille
reaction was selected as a complementary approach to 3
by using Suzuki reactions described earlier.5 Coupling
of organostannanes to â-halo-unsaturated carbonyl de-
rivatives is precedented in the literature.16
9-cis-Lock ed Retin oid s 3a
a
Reagents and conditions: (i) Pd2(dba)3, AsPh3, NMP, 25 °C, 5
min. (ii) nBuLi, THF, -78 °C, 1 h, 25 °C, 2 h. (iii) MnO2, CH2Cl2,
25 °C. (iv) NaClO2, NaH2PO4, 2-methylbut-2-ene, tBuOH-H2O,
25 °C.
Following the described procedure, dienylstannane 611
was prepared in good yield and excellent stereoselectivity
by stannylcupration of the precursor alkyne with a
“higher order”17 tin cuprate according to the general
procedure of Lipshutz,18 followed by a MeOH quench. On
the other hand, â-bromo-unsaturated aldehydes 510 were
easily obtained from the corresponding cycloalkanones
with PBr3 and DMF. Those aldehydes proved to be highly
TABLE 1. Yield s (%) for th e Tr a n sfor m a tion s Dep icted
in Sch em e 2 a n d Over a ll Yield s [%] of th e En tir e
Sequ en ce fr om Com m er cia l Sta r tin g Ma ter ia ls
trienal 7
retinol 8
retinal 9
retinoic acid 3
7a (95%)
7b (97%)
7c (74%)
7d (80%)
8a (55%)
8b (59%)
8c (69%)
8d (73%)
15a (99%)
15b (99%)
15c (99%)
15d (85%)
3a (80%); [25%]
3b (83%); [24%]
3c (81%); [20%]
3d (77%); [11%)
(12) Dawson, M. I.; Hobbs, P. D.; Chan, R. L.-S.; Chao, W.-R. J . Med.
Chem. 1981, 24, 1214.
(13) (a) de Lera, A. R.; Torrado, A.; Iglesias, B.; Lo´pez, S. Tetrahe-
dron Lett. 1992, 33, 6205. (b) Torrado, A.; Iglesias, B.; Lo´pez, S.; de
Lera, A. R. Tetrahedron 1995, 51, 2435. (c) Torrado, A.; Lo´pez, S.;
Alvarez, R.; de Lera, A. R. Synthesis 1995, 285. (d) de Lera, A. R.;
Iglesias, B.; Rodr´ıguez, J .; Alvarez, S.; Lo´pez, S.; Villanueva, X.; Padro´s,
E. J . Am. Chem. Soc. 1995, 117, 8220. (e) Dom´ınguez, B.; Iglesias, B.;
de Lera, A. R. Tetrahedron 1999, 55, 15071. (f) Dom´ınguez, B.; Iglesias,
B.; de Lera, A. R. J . Org. Chem. 1998, 63, 4135. (g) Dom´ınguez, B.;
Pazos, Y.; de Lera, A. R. J . Org. Chem. 2000, 65, 5917. (h) Pazos, Y.;
Iglesias, B.; de Lera, A. R. J . Org. Chem. 2001, 66, 8483.
(14) For reviews, see: (a) Suzuki, A. Acc. Chem. Res. 1982, 15, 178.
(b) Suzuki, A. Pure Appl. Chem. 1985, 57, 1749. (c) Suzuki, A. Pure
Appl. Chem. 1986, 58, 629. (d) Suzuki, A. Pure Appl. Chem. 1991, 63,
419. (e) Martin, A.; Yang, Y. Acta Chem. Scand. 1993, 47, 221. (f)
Suzuki, A. Pure Appl. Chem. 1994, 66, 213. (g) Miyaura, N.; Suzuki,
A. Chem. Rev. 1995, 95, 2457. (g) Suzuki, A. In Metal-Catalyzed Cross-
Coupling Reactions; Diederich, F., Stang, P. J ., Eds; Wiley-VCH: New
York, 1998; Chapter 2, p 167. (h) Suzuki, A. J . Organomet. Chem. 1999,
576, 147.
(15) For reviews, see: (a) Stille, J . K. Pure Appl. Chem. 1985, 57,
1771. (b) Stille, J . K. Angew. Chem. Int. Ed. Engl. 1986, 25, 508. (c)
Mitchell, T. N. Synthesis 1992, 803. (d) Sato, T. Comprehensive
Organometallic Chemistry II; Abel, E. W., Stone, F. G. A., Wilkinson,
G., Eds.; Elsevier: Oxford, UK, 1995; Vol. 11, Chapter 8, p 355. (e)
Farina, V.; Roth, G. P. In Advances in Metal-Organic Chemistry;
Libeskind, L. S., Ed.; J AI Press: New York, 1996; Vol. 5, p 1. (f) Farina,
V. Pure Appl. Chem. 1996, 68, 73. (g) Farina, V.; Krishnamurthy, V.;
Scott, W. J . The Stille Reaction. Org. React. 1997, 50, 1. (h) Stanforth,
S. P. Tetrahedron 1998, 54, 263. (i) Mitchell, N. T. In Metal-Catalyzed
Cross-Coupling Reactions; Diederich, F., Stang, P. J ., Eds.; Wiley-
VCH: New York, 1998; Chapter 4, p 167. (j) Farina, V.; Krishnamur-
thy, V. The Stille Reaction; Wiley: New York, 1999. (k) Duncton, M.
A. J .; Pattenden, G. J . Chem. Soc., Perkin Trans. 1 1999, 1235.
(16) (a) Stille, J . K.; Sweet, M. P. Tetrahedron Lett. 1989, 30, 3645.
(b) Stille, J . K.; Sweet, M. P. Organometallics 1990, 9, 3189. (c) Elsley,
D. A.; MacLeod, D.; Miller, J . A.; Quayle, P. A Tetrahedron Lett. 1992,
33, 409.
unstable, and were used immediately after their purifica-
tion by distillation.
For the coupling process, we selected the recent
modifications of the Stille protocol introduced by Farina,
a combination of “ligandless” palladium catalyst Pd2(dba)3
and the “soft” ligand AsPh3 in NMP, which accelerates
the coupling of organostannanes and electrophiles.19 For
the coupling of dienylstannane 6 and 2-bromocycloalkene
carbaldehydes 5, 5% Pd2(dba)3 and 5% AsPh3 in NMP
was employed, affording compounds 7 in high yields at
room temperature in very short reaction times (1-5 min)
(Scheme 2).20
Wittig reaction (Scheme 2) of conjugated aldehydes
7a -d with excess (3 equiv)21 phosphorane, generated
upon treatment of â-cyclogeranylphosphonium bromide
49 with n-BuLi at -30 °C, provided polyenes 8a -d with
high stereoselectivity as anticipated in the yields shown
in Table 1.
The 9-cis-retinol analogues 8a -d were then unevent-
fully converted to the final carboxylic acids 3a -d through
a two-step oxidation protocol. 9-cis-Retinal analogues
(19) Farina, V.; Krishnan, B. J . Am. Chem. Soc. 1991, 113, 9585.
(20) The reaction is virtually instantaneous at room temperature
regardless of the Pd:ligand ratio (Pd:L ratios of 1:0, 1:0.5, 1:1. 1:2, 1:3,
and 1:4 gave identical rates). We thank Bele´n Vaz from this laboratory
for performing the control experiments.
(21) (a) Mu¨ller, N.; Hoffmann, W. Synthesis 1975, 781. (b) Brooks,
D. W.; Bevinakatti, H. S.; Kennedy, E.; Hathaway, J . J . Org. Chem.
1985, 50, 628. (c) Srikrishna, A.; Krishnan, K. J . Chem. Soc., Perkin
Trans. 1 1993, 667. (d) Sum, F. W.; Weiler, L. Tetrahedron 1981, 37
Suppl. I, 309.
(17) Lipshutz, B. H.; Sengupta, S. Org. React. 1992, 41, 135.
(18) Lipshutz, B. H.; Ellsworth, E. L.; Dimock, S. H.; Reuter, D. C.
Tetrahedron Lett. 1989, 30, 2065.
5878 J . Org. Chem., Vol. 67, No. 17, 2002

Stereoselective Synthesis of Annular 9-cis-Retinoids
TABLE 2. Dissocia tion Con sta n ts a n d Low est
In ter a ction En er gy for th e An n u la r 9-cis-Retin oic Acid
An a logu es
be compared with the experimentally determined dis-
sociation constants (see Table 2).
As seen from this comparison, the experimentally
observed affinity ranking (for the aliphatic rings) abides
by the following ring size order: 3b > 3d > 3a . 3c.
Some of the differences in the experimental binding
free energies are rather small. For instance, the differ-
ences in free energy of binding for 3c and 3d do not
exceed 0.5 kcal/mol. Nevertheless, as seen from the
values of Table 2, the interaction energy reproduces some
of the features in the ranking of the dissociation con-
stants, such as the affinity ranking order for analogues
with an even number of bonds, and most notably the drop
in the affinity shown by 3c with respect to 3b by 1.5 kcal/
mol. This outcome indicates that 3c is not able to
generate an optimal set of interactions with the receptor
as opposed to the other aliphatic analogues. Figure 1
displays the ligand receptor contacts for both the 3b and
3c analogues. The set of residues interacting with the
ligands and the number of van der Waals contacts differ
from one ligand to the other. While analogue 3b interacts
with residues Ala272, Trp305, Asn306, Leu309, and
Leu436, analogue 3c contacts residues Ala272, Trp305,
Asn306, Leu309, Ile310, and Cys 432. Nevertheless, 2c
produces a larger number of van der Waals contacts (19)
with a smaller number of residues (5) than the seven-
member analogue 3c, which contacts 6 residues produc-
ing only a total of 17 van der Waals interactions. The
difference in the number of van der Waals contacts
explains the difference in affinity. The folding of the ring
fragment should play a role in allowing some of the
analogues to optimize their contacts with the receptor.
The only discrepancy between the observed and cal-
culated affinity rankings for the receptor resides in
analogues 3a and 3c. As seen from Table 2, the Kd values
show that the five-membered-ring analogue is a better
binder than the seven-membered-ring analogue, while
the interaction energies indicate the opposite. There may
be several reasons for this disagreement. The results of
the modeling could be affected by the force field used in
our calculations. In the present study we have used
CHARMM, which has been optimized for biomolecules
in general rather than specifically for polyenic com-
pounds. Nevertheless, it has been found that CHARMM-
based molecular mechanics calculations of â-ionone, a
retinoic acid metabolic precursor, reproduce well its
preferred binding conformations to â-lactoglobulin, as
observed in NMR experiments.28
Kd
(nM)
interaction energy
(kcal/mol)
analogue
3a
3b
3c
3d
199
89
1188
146
-222.3
-225.5
-223.9
-224.3
9a -d were acquired after allylic oxidation of alcohols
8a -d with MnO2,22 whereas the desired configurationally
constrained 9-cis-retinoic acid analogues 3a -d were
obtained by oxidation of 9a -d with Lindgrem conditions
(NaClO2, NaH2PO4, 2-methyl-2-butene, t-BuOH-H2O, 25
°C)23 in the yields listed in Table 1.24
RXR Bin d in g Assa y. The assay was performed as
previously described.25 Briefly, COS-7 cells were trans-
fected with a pSG-derived expression vector encoding for
human RXRR with the Polybrene technique. Cells were
lysed, and the nuclei were recovered by centrifugation.
For competition binding assays, nuclear extracts were
incubated with [3H]-9-cis-retinoic acid (50 nM) as the
radioligand and various concentrations of the analogues.
Separation of free and bound ligand was performed by
using an hydroxylapatite gel. The dissociation constant
(Kd value) for each retinoid was determined by nonlinear
regression analysis with use of the Origin software
(Microcalc Software Inc.). Kd values (nM; for comparison
Kd ) 50 nM for 2) are listed in Table 2.
As seen from the values of Table 2, there is no
monotonic variation of Kd with ring size. This points out
the limitation in space near C19 imposed by the receptor.
The best binder is the cyclohexenyl derivative 3b, which
in related studies has been described as a specific RXR
ligand and activator of RXR-mediated pathways.5b
Molecu la r Mod elin g of th e Bin d in g Mod es a n d
Recep tor Affin ities. To rationalize the trends shown
in Table 2, Molecular Modeling-based docking studies
were carried out. As detailed in the Experimental Section,
a large number of conformers were generated and docked
into the receptor to determine their interaction ener-
gies.26,27 This procedure allows the determination of a
“preferred” binding conformation, the one having the
lowest interaction energy. Table 2 lists the interaction
energy between the receptor and the ligand in its
preferred conformation for every analogue, which should
Analysis of the unrestrained Molecular Dynamics (MD)
simulations of the analogues at 300 K gives some clues
about the source of the remaining disparity between the
observed and calculated ranking of componds 3a and 3c.
A possible rationale for this discrepancy can be found in
the flexibility of both ligands in the unbound state as
characterized by the root-mean-square (RMSD) fluctua-
tions with respect to the average structure of a given
analogue. Figure 2 depicts the time evolution of the
RMSD fluctuations for both the five- (3a ) and seven- (3c)
membered-ring 9-cis-locked retinoids. The analysis was
performed on 100 ps of MD trajectory and every frame
was taken at every ps of this trajectory. As seen from
(22) (a) Fatiady, A. J . Synthesis 1976, 65. (b) Broek, A. D.; Muradin-
Szweykowska, M.; Courtin, J . M. L.; Lugtenburg, J . Recl. Trav. Chim.
Pays-Bas 1983, 102, 46.
(23) (a) Lindgrem, B. O.; Nilsson, T. Acta Chem. Scand. 1973, 27,
888. (b) Kraus, G. A.; Taschner, M. J . J . Org. Chem. 1980, 45, 1175.
(c) Bal, B. S.; Childers, W. E.; Pinnick, H. W. Tetrahedron 1981, 37,
2091.
(24) A related analogue incorporating a ring bridging C8 and C11
has been recently described by Ito, which besides the 9Z-configuration,
locks the C8-C9 double bond in an s-trans conformation. (a) Katsuta,
Y.; Aoyama, Y.; Osone, H.; Wada, A.; Uchiyama, S.; Kitamoto, T.;
Masushige, S.; Kato, S.; Ito, M. Chem. Pharm. Bull. 1994, 42, 2659.
(b) Katsuta, Y.; Aoyama, Y.; Osone, H.; Wada, A.; Ito, M. J . Chem.
Soc., Perkin Trans. 1 1997, 405.
(25) Martin, B.; Bernardon, J . M.; Cavey, M. T.; Bernard, B.;
Carlavan, I.; Charpentier, B.; Pilgrim, W. R.; Shroot, B.; Reichert, U.
Skin Pharmacol. 1992, 5, 57.
(26) InsightII, Builder, and Biopolymer are trademarks of MSI, Inc.
San Diego, CA (U.S.A.).
(27) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;
Swaminathan, S.; Karplus, M. J . Comput. Chem. 1983, 4, 187.
(28) Curley, R. W., J r.; Sundaram A. K.; Fowble, J . W.; Abildgaard
F.; Westler, W. M.; Markley, J . L. Pharm. Res. 1999, 16, 651.
J . Org. Chem, Vol. 67, No. 17, 2002 5879

Otero et al.
F IGURE 1. Close-up views of the van der Waals interactions (up to 3.8 Å) of the ring-heavy atoms (green) with the receptor
(yellow) for analogues 3b (upper panel) and 3c (lower panel).
this figure, the RMSD fluctuations are larger for 3c than
for 3a , in line with the larger number of degrees of
freedom available to the former compound. These results
indicate that the entropy penalty for binding 3c to RXRR
is larger than that for binding 3a . Including a confor-
mational entropy term in the prediction of the binding
free energy could improve the agreement between the
observed and calculated binding ranking of these com-
pounds.
densation reactions, followed by adjustment of the oxida-
tion state. The binding affinity of these compounds has
been determined and they have been rationalized by
analysis of Molecular Modeling based docking of these
compounds to the receptor. Moreover, the alicyclic bind-
ing rank was shown to be determined to a large extent
by their interaction with the receptor. The outcome
indicates that the available space is limited and it is not
the only factor determining the binding capabilities.
Including entropic terms that take into account the larger
flexibility of those analogues with larger aliphatic rings
could help to bridge the only remaining difference
In summary, configurationally constrained 9-cis-ret-
inoic acid analogues have been efficiently prepared by
using a combination of Stille coupling and Wittig con-
5880 J . Org. Chem., Vol. 67, No. 17, 2002

Stereoselective Synthesis of Annular 9-cis-Retinoids
(2E,4E)-3-Meth yl-5-{2-[(E)-2-(2,6,6-tr im eth ylcycloh ex-
1-en -1-yl)eth en yl]-1-cycloh exen -1-yl}-2,4-pen tadien al (9b).
To a mixture of alcohol 8b (34 mg, 0.10 mmol) in CH2Cl2 (3
mL) was added MnO2 (181 mg, 2.08 mmol) and the resulting
suspension was stirred at room temperature for 16 h. The
reaction mixture was filtered through Celite and the solvent
was removed in vacuo to afford, after chromatography, retinal
9b (33 mg, 99%) as a pale orange oil. 1H NMR (CDCl3, 250
MHz) δ 1.04 (6H, s), 1.4-1.8 (8H, m), 1.76 (3H, s), 2.04 (2H, t,
J ) 5.9 Hz), 2.33 (3H, s), 2.3-2.4 (4H, m), 6.00 (1H, d, J ) 8.2
Hz), 6.27 (1H, d, J ) 16.1 Hz), 6.35 (1H, d, J ) 15.7 Hz), 6.70
(1H, d, J ) 16.1 Hz), 7.47 (1H, d, J ) 15.7 Hz), 10.10 (1H, d,
J ) 8.2 Hz); 13C NMR (CDCl3, 63 MHz) δ 13.1, 19.2, 21.8, 22.3
(2×), 26.3, 27.2, 29.0 (2×), 33.0, 34.2, 39.5, 128.9, 129.1 (2×),
129.8, 130.2, 131.0, 133.6, 138.3, 139.4, 155.6, 191.2; IR (CH2-
Cl2) υ 1726 (s, CdO) cm-1; UV (MeOH) λmax 370 nm; MS m/z
(%) 324 (M+, 100), 309 (18), 291 (14); HRMS [M+] calcd for
F IGURE 2. RMSD fluctuations of the structures of com-
pounds 3a (thin line) and 3c (thick line) with respect to their
averaged structure in an MD simulation, as a function of time.
C
23H32O 324.2453, found 324.2451.
(2E,4E)-3-Meth yl-5-{2-[(E)-2-(2,6,6-tr im eth ylcycloh ex-
1-en -1-yl)et h en yl]-1-cycloh exen -1-yl}-2,4-p en t a d ien oic
Acid (3b). To a solution of aldehyde 9b (38 mg, 0.12 mmol)
and 2-methyl-2-butene (0.64 mL, 5.90 mmol) in t-BuOH (3 mL)
was added, using a syringe pump, for a period of 73 min, 0.6
mL of a solution of NaClO2 (67 mg, 0.74 mmol) and NaH2PO4
(53 mg, 0.44 mmol) in H2O. The reaction mixture was stirred
at room temperature for 10 h and then the pH was raised to
pH ∼10 by addition of 3 M NaOH. The t-BuOH was evaporated
under vacuum and the remaining residue was diluted with
water, saturated with NaCl, and extracted with hexane. The
aqueous layer was acidified to pH ∼3 with 0.5 N HCl and then
extracted with ether (3×). The combined organic layers were
dried over MgSO4, filtered, and concentrated. The residue was
purified by chromatography (silica, 70:30 hexane/ethyl acetate)
to afford retinoic acid 3b (33 mg, 83%) as a yellow solid (mp
163-164 °C; lit.5a mp 165-167 °C). 1H NMR (CDCl3, 400 MHz)
δ 1.04 (6H, s), 1.5-1.7 (8H, m), 1.76 (3H, s), 2.0-2.1 (2H, m),
2.37 (3H, s), 2.3-2.4 (4H, m), 5.84 (1H, s), 6.25 (1H, d, J )
15.6 Hz), 6.32 (1H, d, J ) 16.1 Hz), 6.72 (1H, d, J ) 15.6 Hz),
7.39 (1H, d, J ) 16.1 Hz); 13C NMR (CDCl3, 100 MHz) δ 14.0,
19.2, 21.8 (2×), 22.4, 26.3, 27.1 (2×), 29.0 (2×), 33.0, 34.2, 39.4,
117.2, 128.2, 129.4, 129.8, 130.2, 130.8, 132.8, 138.1, 138.2,
156.0, 171.3; IR (CH2Cl2) υ 1682 (m, CdO) cm-1; UV (MeOH)
λmax 266, 336 nm; MS m/z (%) 340 (M+, 100), 325 (14); HRMS
[M+] calcd for C23H32O2 340.2402, found 340.2395.
between the observed dissociation constants and the
predicted affinities.
Exp er im en ta l P a r t
Gen er a l Exp er im en ta l P r oced u r es: 2-[(1E,3E)-5-Hy-
d r oxy-3-m eth ylp en ta -1,3-d ien -1-yl]cycloh ex-1-en -ca r ba l-
d eh yd e (7b). To a solution of â-bromoaldehyde 5b (1.0 g, 5.29
mmol) in NMP (15 mL) were added AsPh3 (65 mg, 0.21 mmol)
and Pd2(dba)3 (194 mg, 0.21 mmol). After the mixture was
stirred at room temperature for 10 min, a solution of vinyl-
stannane 6 (2.37 g, 6.13 mmol) in NMP (24 mL) was added,
and the reaction mixture was again stirred at room temper-
ature for 5 min. The reaction was quenched by addition of a
saturated aqueous KF solution (20 mL) and stirring was
maintained for 30 min. The aqueous layer was extracted with
ether (3×) and the combined organic layers were washed with
saturated KF (3×) and H2O (3×), dried (MgSO4), filtered, and
concentrated. The residue was purified by chromatography
(silica, 70:30 hexane/ethyl acetate) to yield 1.06 g (97%) of
aldehyde 7b as an orange oil. 1H NMR (CDCl3, 400 MHz) δ
1.5-1.6 (4H, m), 1.77 (3H, s), 2.22 (2H, t, J ) 6.0 Hz), 2.37
(2H, t, J ) 6.0 Hz), 4.26 (2H, d, J ) 6.7 Hz), 5.74 (1H, t, J )
6.7 Hz), 6.45 (1H, d, J ) 15.7 Hz), 7.10 (1H, d, J ) 15.7 Hz),
10.26 (1H, s); 13C NMR (CDCl3, 100 MHz) δ 12.6, 21.4, 21.9,
23.1, 27.5, 59.4, 122.6, 134.1, 135.2, 135.7, 137.6, 152.1, 190.6;
Full details of the synthesis and spectroscopic characteriza-
tion of analogues 3a , 3c, and 3d are described in the Support-
ing Information.
IR (CH2Cl2) υ 3600-3100 (broad, OH), 1656 (s, CdO) cm-1
;
UV (MeOH) λmax 238, 312 nm; MS m/z (%) 206 (M+, 4), 197
(3); HRMS [M+] calcd for C13H18O2 206.1307, found 206.1307.
(2E,4E)-3-Meth yl-5-[2-[(E)-2-(2,6,6-tr im eth ylcycloh ex-
1-en -1-yl)et h en yl]-1-cycloh exen -1-yl}-2,4-p en t a d ien -1-
ol (8b). To a cooled (-30 °C) suspension of phosphonium salt
4 (1.52 g, 3.16 mmol) in THF (7 mL) was added n-BuLi (2.57
mL, 1.23 M in THF, 3.16 mmol) and the resulting solution
was stirred at 0 °C for 1 h. Aldehyde 7b (0.18 g, 0.87 mmol) in
THF (3 mL) was then added via cannula at -78 °C, and the
reaction mixture was stirred at -78 °C for 1 h and at 25 °C
for 2 h. Water was added, and the aqueous layer was extracted
with ether (3×). The combined organic layers were washed
with H2O (3×) and brine, dried (MgSO4), and concentrated.
Purification by chromatography (silica, 80:20 hexane/ethyl
Com p u ta tion a l Meth od s
P r ep a r a tion of Molecu la r System s for th e Ca lcu la -
tion s. The starting point for the calculations was the recently
obtained structure of the RXRR receptor bound to its cognate
ligand 9-cis-retinoic acid (PDB code 1FYB). The hydrogens
were added to this complex by using the Biopolymer and
Builder modules in the InsightII molecular display and
handling package, at pH 7 and with the N and C end caps
charged.26
Modelin g th e 9-cis-Retin oic An alogu es. Using the Builder
module in the MSI software package,26 the 5-, 6-, 7-, and
8-membered alicyclic rings were created between atoms C19
and C10 of the molecular backbone of this molecule (see
Scheme 1 for atom numbering). To generate multiple confor-
mations of the aliphatic rings in these molecules, a molecular
dynamic (MD) trajectory (100 ps in length) at 1000 K in an
NVT ensemble was run, keeping most of the backbone atoms
rigid, except for atoms C10 and C19. This MD simulation
produced 100 conformations for each molecule, spaced at 1 ps.
Additionally, 100 ps of a MD simulation, at 300 K (in a NVE
ensemble) with all conformational internal degrees of freedom
unconstrained in a vacuum, was run to determine the confor-
mational phase space span by every ligand.
1
acetate) afforded 0.17 g (59%) of retinol 8b as a yellow oil. H
NMR (CDCl3, 250 MHz) δ 1.03 (6H, s), 1.4-1.8 (8H, m), 1.74
(3H, s), 1.86 (3H, s), 1.9-2.0 (2H, m), 2.3-2.4 (4H, m), 4.29
(2H, d, J ) 7.0 Hz), 5.70 (1H, d, J ) 7.0 Hz), 6.14 (1H, d, J )
16.1 Hz), 6.28 (1H, d, J ) 15.7 Hz), 6.70 (1H, d, J ) 16.1 Hz),
6.95 (1H, d, J ) 15.7 Hz); 13C NMR (CDCl3, 63 MHz) δ 12.6,
19.2, 21.8, 22.6 (2×), 26.5, 26.8, 28.9 (2×), 33.0, 34.2, 39.5, 59.5,
126.4, 126.5, 128.9, 129.6, 130.8 (2×), 131.3, 134.1, 137.4,
138.5; IR (CH2Cl2) υ 3600-3100 (broad, OH) cm-1; UV (MeOH)
λ
max 240, 282 nm; MS m/z (%) 326 (M+, 100), 311 (8), 295 (32);
HRMS [M+] calcd for C23H34O 326.2610, found 326.2603.
J . Org. Chem, Vol. 67, No. 17, 2002 5881

Otero et al.
Estim a tin g th e Bin d in g Affin ities. Every single confor-
mation of the 9-cis-retinoic derivatives obtained with the
restrained MD was docked to the nuclear receptor keeping the
same conformation as the original ligand by using the tools of
the InsightII molecular display and handling software. The
resulting complexes were subjected to a 1000-step energy
minimization, using the steepest descent protocol and the
CHARMM force field.27 To rank the binding affinity of each of
the 100 conformations of every ligand, we calculated the
interaction energy between ligand and receptor using the same
force field. This protocol allowed the prediction of the “best
binding” conformation for every molecule, which was compared
with the experimentally determined affinities.
Xunta de Galicia (Grant PGIDT99PXI30105B, Visiting
Research Contract to F. Sussman, and fellowship to M.
P. Otero), and Galderma R & D for financial support.
We wish to thank Drs. Uwe Reichert, Philippe Diaz, and
Serge Michel (Galderma R & D) for the RXRR-binding
studies.
Su p p or tin g In for m a tion Ava ila ble: Experimental pro-
cedures for the synthesis of 3a , 3c, and 3d and 1H and 13C
NMR spectra for compounds described in the Experimental
Section. This material is available free of charge via the
Internet at http://pubs.acs.org.
Ack n ow led gm en t . We thank DGICYT (Grant
SAF98-0134, which also supported Dr. A. Torrado),
J O0257391
5882 J . Org. Chem., Vol. 67, No. 17, 2002