Stereoselective synthesis of new digitalis-like perhydroindene derivatives from the Hajos-Parrish ketone

Article abstract of DOI:10.1055/s-1998-1902

New cardenolide-like perhydroindene derivatives 7a,b and 8 were synthesized from (+)-(3aS,7aS)-3a-hydroxy-7a-methylperhydroinden-1,5-dione 1 (Hajos-Parrish ketone) in few stereo- and regioselective steps and without using protective groups.

Full text of DOI:10.1055/s-1998-1902

1234
LETTERS
SYNLETT
Stereoselective Synthesis of New Digitalis-like Perhydroindene Derivatives from the Hajos-
Parrish Ketone
Nicoletta Almirante,* Alberto Cerri, Sergio De Munari
Prassis Istituto di Ricerche Sigma-Tau, Via Forlanini 3, 20019 Settimo Milanese, (MI), Italy
Fax:+39 02 33500408; E-mail: mc3405@mclink.it
Received 22 July 1998
Abstract: New cardenolide-like perhydroindene derivatives 7a,b and 8
were synthesized from (+)-(3aS,7aS)-3a-hydroxy-7a-methyl-
perhydroinden-1,5-dione 1 (Hajos-Parrish ketone) in few stereo- and
b)
c)
1α-substituted derivatives are more stable than 1β-derivatives
(like 17α- vs. 17β- in the 14β-androstane series);
the flexible two-rings system can assume two different con-
formations depending on the stereochemistry of the substituents.
In particular, in 1-oxo or 1β-derivatives, a substituent at position 5
stabilizes largely the conformation in which it is equatorial, as
confirmed by molecular mechanics calculations (MM2). In the
case of 5α-substituted derivatives this leads to the predominance
of the non-digitalis-like conformation (see Fig. 2).
regioselective steps and without using protective groups.
Two hundred years since their introduction into therapy, digitalis
glycosides still remain in a prominent position for the treatment of
congestive heart failure, despite their unfavourable therapeutic ratio.1
In the light of this, the synthesis of analogues possessing different
functional groups and/or altered ring systems acquires special value in
the search for less toxic compounds.
In the steroidal moiety of the cardiac glycosides (cardenolides) the C/D
cis ring junction, the 17β-butenolide moiety and the 14β-hydroxyl
function are recognized as three specific features for a potent
pharmacological action. On the other hand, the A/B ring junction can
vary from cis to trans (e.g. digitoxigenin and uzarigenin, Fig. 1) without
a dramatic loss of activity. As an ongoing project aimed at searching for
new digitalis-like compounds with an improved pharmacological
profile, we decided to synthesize a new class of derivatives in which the
C/D part of cardenolides is maintained while the B ring is abolished to
give simplified 5β-aryl or 5β-cyclohexylperhydroindene derivatives:
7a,b and 8 (Fig. 1) are the leading compounds of this new class.
Scheme 1
Figure 1
The retrosynthetic analysis for products 7a,b and 8 is shown in Scheme
1: the Hajos-Parrish ketone 12 emerged as a useful chiral building block
possessing at C-3a and C-7a the same configuration of the natural
compounds. While 1 is the starting material for many syntheses of
vitamin D3 analogues (trans perhydroindene derivatives), to our
knowledge it has seldom been used for cardenolide construction and
only one-carbon elongations at C-1, after protection3 or derivatization4
of the 5-keto function, were reported.5 The main reasons are probably as
follows:
Figure 2
With these considerations in mind, our approach was to introduce the
5β-substituent in the first step to afford a perhydroindene derivative that
could be conformationally related to a 14β-androstane. This compound
was therefore expected to be easily functionalized at the 1β-position by
a free-radical stereoselective reaction, as shown by Daniewsky6a and
a)
as for the 14β-androstane series, the cis ring junction is the cause
for the weak reactivity of the 1-keto function;

November 1998
SYNLETT
1235
Wicha6b for other C/D cis steroids. On the basis of our previous
experience with the digitalis skeleton,7 maleic anhydride was preferred
to tert-butyl isocyanide as radical acceptor, being a more advanced
precursor of the butenolide moiety.8
The straightforward route to the synthesis of 7a,b and enantiomerically
pure 8 starting from 1 is reported in Scheme 2. The 5β-substituent was
first introduced by reaction of 1 with PhLi (3 equiv., -78 °C, THF). The
reaction was regioselective, but produced a 1:1 mixture of 5β- and 5α-
phenyl derivatives (2 and 5-epi-2) and the yield was a scanty 40%,
probably due to enolization of the ketone (Scheme 3).
To circumvent the enolization problem we repeated the arylation on the
Scheme 3
CeCl3/C=O complex9 (CeCl3
1 equiv., THF, room temperature,
overnight, then -78°C, PhLi 3 equiv.). As expected, the yield increased
to a satisfactory 73% and, moreover, the β/α isomeric ratio was 4.6:1.
No attempts were made to improve this ratio since the two
diastereoisomers were easily separated by flash chromatography.
1
signal (W1/2h = 26.2 Hz, axial) in the H NMR spectra of the reduced
compounds (3a and 5-epi-3a).
The desired cyclohexyl derivative 3b was obtained by hydrogenation of
the aromatic ring of 3a (MeOH, H2/Rh/Al2O3, 45 psi, 4 h). The 1-keto
derivatives 3a,b were transformed into the 1α-iodo compounds 4a,b by
a stereospecific reaction sequence, which had already been applied to
CD-cis steroids.11 Free-radical reaction with maleic anhydride (maleic
anhydride 2 equiv., TTMSS 2.2 equiv., toluene, 90 °C, 1 hr, then DBU
2.2 equiv. diethyl ether, 0 °C to room temperature, overnight) afforded
1β-substituted 6a,b,12 probably through the unisolated anhydrides 5a,b.
Notably, 1α-isomers were not detected, therefore the free radical
reaction was completely stereospecific also on this more flexible
nucleus.
The benzylic 5α-hydroxy group of 2 was selectively hydrogenolyzed by
excess Raney-Nickel (EtOH/H2O, reflux, 4 h) affording 3a, with
complete retention of configuration.10 Analogously 5-epi-3a was
obtained from 5-epi-2, for spectroscopic comparison. The configuration
of compounds 2 and 5-epi-2, 3a and 5-epi-3a, were established by 13C
NMR analysis of the isomeric pairs in comparison with the spectrum of
3β,14-dihydroxy-5β,14β-androstan-17-one. In fact, the C-18 methyl
resonance of the latter compound is found at 12.8 ppm as in the case of
compounds 2 and 3a. On the contrary, the methyl group of their epimers
resonates at ca. 19.4 ppm, thus indicating
a non-digitalis-like
Chemoselective reduction of the ester function13 of 6a,b (NaBH4/
MeOH, THF reflux, 1hr, then HCl, pH 1.4 overnight) led to the
butanolide derivatives 7a,b.14 The transformation of compounds 7b into
the final compound 8 was achieved through well-known selenoxide
conformation for the 5α-isomers. In addition, evidence for a strong
hydrogen bonding is found only in the 1H NMR spectrum of 5-epi-2
where the two hydroxyl groups are 1,3-diaxial. Finally, the equatorial
position of the phenyl group in both epimers is confirmed by the H-5
Scheme 2

1236
LETTERS
SYNLETT
chemistry (THF, LDA 6 equiv., -20 to 0 °C, 1 hr; -78 °C, PhSeCl 6.2
equiv., 1.5 h; quenching at pH 3.5; THF, AcOH, -10 °C, 30% H2O2 10
equiv.) in 50% yield. Other higher-yielding reported methods, involving
first the preparation of the trimethylsilyl ketene acetal derivative of 7b
and prevented the dehydration of the acid labile 3aβ-hydroxy
function .
(13) Soai, K.; Oyamada, H. Synthesis 1984, 605.
(14) For the critical workup of this reaction see ref 7. Compounds 7a,b
were isolated, each one, as an almost 55:45 diastereoisomeric
mixture.
15
followed by either a) oxidation with Pd(OAc)2 or b) TMSOTf
mediated sulfenylation and elimination16 led in the first step only to
silylated open chain derivatives.
(15) Mukai, C.; Moharram, S. M.; Azukizawa, S.; Hanaoka, M., J.
Org. Chem.. 1997, 62, 8095.
In conclusion, the simplified cardanolides 7a,b and the cardenolide 8
with a perhydroindene skeleton, were obtained in 14% and 7% final
yields, respectively from the known, enantiopure compound 1 through a
simple and versatile reaction sequence.17 The key steps were the
introduction of a phenyl substituent at 5β-position and then of the
butenolide moiety at 1β-position. The two new stereogenic centers were
established by means of few regio- and stereoselective reactions,
avoiding protection-deprotection steps.
(16) Wilson, L. J.; Liotta, D. C. J. Org. Chem. 1992, 57, 1948.
(17) All new compounds gave spectroscopic data in agreement with
the assigned structures. Selected data follow:
2: 1H NMR (300 MHz, CDCl3): δ 3.09 (m, 1H), 1.13 (s, 3H); 13
C
NMR (75 MHz): δ 221.3, 79.4 (C-3a), 73.9 (C-5), 52.2 (C-7a),
12.7 (CH3);
1
5-epi-2: H NMR (300 MHz, CDCl3): δ 4.29 (bs, 1H, OH), 2.62
Compound
8 showed a low biological activity, evaluated as
(bs, 1H, OH), 2.52 (m, 1H), 1.09 (s, 3H); 13C NMR (75 MHz): δ
218.7, 77.6 (C-3a), 74.5 (C-5), 52.4 (C-7a), 19.3 (CH3);
3a: mp 167-169 °C (acetone/water); 1H NMR (300 MHz, CDCl3):
δ 2.79 (m, 1H), 1.19 (s, 3H); 13C NMR (75 MHz): δ 221.0, 80.1
(C-3a), 52.6 (C-7a), 41.8 (C-5), 12.9 (CH3). Calcd for C16H20O2:
C, 78.65; H, 8.25. Found: C, 78.76; H, 8.32.
displacement of the specific 3H-ouabain binding to Na+,K+-ATPase18
and as inhibition of the activity of the same enzyme,19 which is involved
as the first event in the generation of positive inotropism caused by
digitalis compounds. At 10-5 M, 8 showed 30% displacement and 0%
inhibition.
3b; white foam; 1H NMR (300 MHz, CDCl3): δ 1.05 (s, 3H); 13
C
References and Notes
NMR (75 MHz): δ 221.3, 80.4 (C-3a), 53.1 (C-7a), 12.7 (CH3).
Calcd for C16H26O2: C, 76.75; H, 10.47. Found: C, 76.60; 10.44.
4a: white foam; 1H NMR (300 MHz, CDCl3): δ 4.48 (t, J=9.1 Hz,
1H, H-1), 2.65 (m, 1H, H-5), 1.08 (s, 3H); MS, m/z: 229 (M+-I).
Calcd for C16H21IO: C, 53.94; H, 5.94; I, 35.62. Found: C, 53.80;
H, 5.96; I, 35.42
(1) Hoffmann, B. F.; Bigger, J. T. Jr. In The Pharmacological Basis
of Therapeutics; Goodman Gilman, A.; Rall, T. W.; Nies, A. S.;
Taylor, P., Eds; Pergamon Press Inc., 1990; p. 814.
(2) Hajos, Z. G.; Parrish, D. R. J. Org. Chem. 1974, 39, 1615.
(3) Medarde, M.; Tomé, F.; López, J.L.; Caballero, E.; Boya, M.;
1
4b: colourless oil; H NMR (300 MHz, CDCl3): δ 4.43 (t, J=9.2
Melero, C.P.; San Feliciano, A. Tetrahedron Lett. 1994, 35, 8683.
Hz, 1H, H-1), 0.94 (s, 3H); 13C NMR (75 MHz): δ 79.0 (C-3a),
47.8 (C-7a), 16.2 (CH3); MS, m/z (rel intensity): 361 (M+-1, 0.2),
217 (M+-H2O-I, 100). Calcd for C16H27IO: C, 53.04; H, 7.51; I,
35.03. Found: C, 52.83; H, 7.53; I, 34.85.
(4) Frigerio, M.; Santagostino, M.; Sputore, S. Synlett 1997, 833.
(5) While we were submitting this manuscript, two papers, reporting a
synthetic approach to complex cardenolides starting from the
3a,4-anhydro derivative of 1, appeared in the literature: a)
Overman, L.E.; Rucker, P.V. Tetrahedron Lett. 1998, 39, 4643; b)
Hynes, J.H. Jr; Overman, L.E.; Nasser, T.; Rucker, P.V.
Tetrahedron Lett. 1998, 39, 4647
6a: 1H NMR (300 MHz, CDCl3): δ 3.48 (m, 1H), 3.28-3.05 (m,
3H), 2.88-2.60 (m, 2H), 2.48 (m, 2H), 2.30 (m, 1H), 1.28 (s, 3H),
1.15 (s, 3H); 13C NMR (75 MHz): 177.1, 177.0 174.0, 173.6,
92.4, 91.9 (C-3a), 47.3, 47.2 (C-1), 44.4, 43.0 (C-7a), 46.6, 41.5
(CHCOO), 40.9, 40.8 (C-5), 15.8, 14.6 (CH3); MS, m/z: 328 (M+).
Calcd for C20H24O4: C, 73.14; H, 7.31. Found: C, 72.73; H, 7.40.
(6) a) Daniewski, A. R.; Kabat, M. M.; Masnyk, M.; Wicha, J.;
Wojciechowska, W.; Duddeck, H, J. Org. Chem. 1988, 53, 4855.
b) Daniewski, A. R.; Kabat, M. M.; Masnyk, M.; Wojciechowska,
W; Wicha, J., Collect. Czech. Chem. Commun. 1991, 56, 1064.
1
6b: H NMR (300 MHz, CDCl3): δ 3.42 (m, 1H), 3.22-2.97 (m,
3H), 2.62 (dd, J=13, 9 Hz, 1H), 2.45 (m, dd J=16, 7 Hz, 1H), 1.15
(s, 3H), 1.05 (s, 3H); 13C NMR (75 MHz): 176.2 176.4, 174.3,
174.1, 93.3, 92.6 (C-3a), 47.4, 47.3 (C-1), 44.6, 44.2 (C-7a), 46.7,
41.4 (CHCOO), 15.7, 14.4 (CH3); MS, m/z (rel intensity): 335
(M++1, 3), 194 (100). Calcd for C20H30O4: C, 71.82; H, 9.04.
Found: C, 71.55; H, 9.11.
(7) Almirante, N.; Cerri, A. J. Org. Chem. 1997, 62, 3402.
(8) For a higher-yielding conversion of the 17β-CN group into the
butenolide moiety see Stork, G.; West, F.; Lee, H. Y; Isaacs, R. C.
A.; Manabe, S., J. Am. Chem. Soc. 1996, 118, 10660.
1
(9) Imamoto, T.; Kusumoto, T.; Tawarayama, Y.; Mita, T.; Hatanaka,
7a: H NMR (300 MHz, CDCl3): δ 4.51 (t, J=8 Hz, 1H), 4.42 (t,
Y.; Yokoyama, M. J. Org. Chem. 1984, 49, 3904.
J=8 Hz, 1H), 4.04 (t, J=8 Hz, 1H), 3.92 (t, J=8 Hz, 1H), 1.08 (s,
3H), 1.05 (s, 3H);
7b: 1H NMR (300 MHz, CDCl3): δ 4.51 (t, J=8 Hz, 1H), 4.42 (t,
J=8 Hz, 1H), 4.04 (t, J=8 Hz, 1H), 3.89 (t, J=8 Hz, 1H), 0.98 (s,
3H), 0.93 (s, 3H);
8: mp 200-203 °C (EtOH); 1H NMR (300 MHz, CD3OD+
CDCl3): δ 5.90 (m, 1H), 5.05 (d, J=18 Hz, 1H), 4.88 (d, J=18 Hz,
1H), 2.83 (m, 1H), 0.87 (s, 3H); 13C NMR (75 MHz,
CD3OD+CDCl3): δ 178.1, 177.3, 118.0, 83.7 (C-3a), 75.4
(CH2O), 16.8 (CH3); MS, m/z: 318 (M+). Calcd. for C20H30O3: C,
75.43; H, 9.50. Found: C, 75.37; H, 9.58.
(10) Also 5-epi-2 can be converted to 3a by hydrogenolysis with Pd/C
10 % -HClO4 in EtOAc with almost complete inversion of
configuration (β/α 85:15, 80% yield).
(11) Compounds 4a,b were obtained by treatment of the crude
hydrazones of 3a,b with iodine and, successively, reduction of the
1-iodo-1,2-didehydro derivatives with diimide, as described for
the 5β,14β-androst-9(11)-ene nucleus, by Daniewski (ref 6a).
(12) The free radical reaction was performed in toluene at 90 °C with
tris(trimethylsilyl)silane and AIBN. The best workup for the
reaction was the Curran procedure with DBU. As described by us
for the 5β,14β-androstane nucleus,7 it allowed the complete
conversion of the intermediate anhydride 5a,b to compounds 6a,b
(each one isolated as an almost 60:40 diastereoisomeric mixture)
(18) Brown, L.; Erdmann, E. Arzneim.-Forsch. 1984, 34, 1314.
(19) Doucet, A.; Hus-Citharel, A.; Morel, F. Am. J. Physiol. 1986, 251,
F851.