Synthetic study toward total synthesis of (±)-germine: Synthesis of (±)-4-methylenegermine

Article abstract of DOI:10.1021/acs.orglett.7b02434

The total synthesis of 4-methylenegermine is described.

Full text of DOI:10.1021/acs.orglett.7b02434

Letter
pubs.acs.org/OrgLett
Synthetic Study toward Total Synthesis of ( )-Germine: Synthesis of
( )-4-Methylenegermine
Gilbert Stork,*,† Ayako Yamashita,† Robert M. Hanson, Ly Phan, Eifion Phillips, Daniel Dube,
́
Pieter H. Bos,‡ Andrew J. Clark, Maxwell Gough, Mark L. Greenlee, Yimin Jiang, Keith Jones,
Masato Kitamura, John Leonard, Tongzhu Liu, Philip J. Parsons, and Aranapakam M. Venkatesan
‡
†Department of Chemistry and Chemical Probe Synthesis Facility, Department of Biological Sciences, Columbia University, New
York, New York 10027, United States
S
* Supporting Information
ABSTRACT: The total synthesis of 4-methylenegermine is
described.
ermine (1, Figure 1)1 is the parent structure of a number of
Scheme 1. Conversion of 3 to Tricyclic Compound 9
Galkaloids that are esters or polyesters of its hydroxyl groups.
Even though germine has been known for quite some time, no
synthetic effort has been reported. There was great interest in the
possible use of this and related substances in controlling blood
pressure.2 Our initial synthetic goal was to construct the germine-
like structure 2 in which all relative stereocenters are introduced
with the correct stereochemistry.3 One extra methylene group on
C4 of the A ring was added to the hemiketal structure to make the
intermediates more stable and to avoid the known sensitivity of
such compounds to bases. We hoped to remove this superfluous
methylene group at a later stage of the synthesis but retain the
established relative stereochemical configuration at C4.
in DCM, at −10 °C, gave a 4:1 mixture of the epoxides 5α and
5β. The major component was isolated and analyzed by X-ray
analysis to reveal the α-epoxide 5α.7 For conversion of 5α to the
desired β-allylic alcohol 6, the mixture (5α + 5β) could be
eliminated and isomerized/equilibrated8 with aluminum iso-
propoxide, [Al(OiPr)3], in boiling m-xylene to provide a 1:4
mixture of the allylic alcohols. The major component was
confirmed as the more stable 3β-hydroxy-4-methylene com-
pound 6 by X-ray analysis.
The small amount of unwanted 5α isomer as well as an isomer
of 6 could be recycled with the next batch of the epoxides. Taking
in account one recycling, the yield for the required conversion of
4 to 6 was 75−80%. Hydroxylation of 6 with osmium tetroxide
(OsO4)9 in the presence of NMO gave a single triol in 81% yield.
It was assumed to be 710 and ready for attempted connection to
ring C. At the time these experiments were done, there had not
been a report of the trapping of an alcohol group by the cation, or
radical cation, formed from the oxidation of an anisole ring. More
Figure 1. Structures of germine (1) and 4-methylenegermine (2).
The starting material chosen for the germine synthesis was the
readily available tricyclic ketone 3.4 We planned to use the A/B
rings of ketone 3 as the precursors of the A/B rings in germine
(Scheme 1). Therefore, the starting enone had to be reduced to
produce the cis A/B fusion. We hoped to do this by applying
Kabalka’s method:5 the reduction of the tosylhydrazone from 3
with catecholborane. This took place to give the tricyclic olefin 46
in 73% yield. Not only was the cis stereochemistry of the A/B
system produced, but the double bond was also placed correctly
for the oxygenation of ring A. Oxidation of 4 with peracetic acid
Received: August 7, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.7b02434
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
recently, examples of this reaction have appeared in the
literature.11 Several oxidations were carried out on 7; some
electrochemical, some with CAN, some on various derivatives of
7, such as the diacetate, the acetonide, the benzylidene, and the
mono- and di-TBS derivatives. In the end, we selected direct
oxidation of 7 with 2.1 equiv of CAN in aq acetonitrile at −20 °C.
This gave, in >90% yield, the crude dihydroxydienone 8 which
was immediately transformed into its acetonide 9 in 71% overall
yield from 6. As noted, the acetonide system of 9 involves a C4
hydroxymethyl group rather than the required hemiketal
hydroxyl of 1.
MeOH−THF) gave a vicinal diol mixture 18 in 90% yield from
15. Cleavage (sodium periodate/aq p-dioxane) of 18 gave the
dialdehyde 19 (Scheme 3).
Scheme 3. Construction of C Ring and Preparation of Diene
22 and Dienophile 23
Now three problems had to be faced: (1) reduction of the
dienone 9 to the enone 10; (2) use of the cyclohexenone system
to introduce the α hydroxyl group at C7; (3) reduction of the
cyclohexenone ring C to form a required trans B/C system, the
precursor of the five-membered ring needed for further
elaboration (Scheme 2).
Scheme 2. Functionalization of Tricyclic System 15
The idea had been to cyclize 19 to the cyclopentene aldehyde
20. There are two possible ways for the dialdehyde to cyclize.
The undesired cyclization would involve attack by base on a
methylene in a 1,3 relationship to the angular methyl group, and
this would favor the alternative methylene, thus leading to the
correct aldol. Surprisingly, the cyclization with piperidine acetate
salt stopped at the aldol state, rather than giving directly the
unsaturated aldehyde. Indeed, dehydration was not simple,
presumably reflecting the added strain a double bond would
place in the trans-hydrindane system: heating at reflux a DCM
solution of the mesylate (from MsCl/Et3N on the aldol) left the
mesylate mixture unchanged. Only by treatment of the mesylate
with DMAP in an acetonitrile−THF mixture at 45−50 °C, for 24
h, was 20 obtained, only in 40−50% yield (overall from 19). This
was a good place to confirm the geometry achieved so far in our
synthesis of 20, and an X-ray structure of the derivative 2115 was
obtained.
The further elaboration of 20 toward 2 involved transforming
it into an alkoxydiene, which could take part in a Diels−Alder
cycloaddition with a suitable dienophile. This was done by Wittig
reaction with (benzyloxymethyl)triphenylphosphorane, easily
prepared by deprotonation of the alkylation product of
triphenylphosphine with chloromethyl benzyl ether to give a
suitable diene 22 in 77% yield. We would have liked to obtain a
preponderance of the E-enol ether from the Wittig reaction, so
that the following Diels−Alder cycloaddition would lead to the
correct hydroxyl stereochemistry, protected as its benzyl ether.
The E/Z ratio of enol ethers expected from a Wittig reaction is,
however, not clear (nor understood). Because such enol ethers
are mostly made as intermediates to the aldehydes obtained by
hydrolysis, characterization of the enol ether geometry is not
usually of concern. In the few cases investigated, both isomers are
formed.16 Separation of the predominately E-enol ether (E/Z =
4:1, in our case 22) was unnecessary, as the Z-isomer would be
expected to undergo the planned Diels−Alder addition with
much more difficulty.
Reduction of the less substituted double bond could be done12
by the use of the Wilkinson catalyst under hydrogen in equal
amounts of benzene and THF. This gave 10 in 85% yield.
Introduction of the 7α-hydroxyl was initially attempted by
peracid treatment of the enol acetate of 10, but that was
unsatisfactory. We then followed a Marshall protocol:13 initial
reduction of the enol acetate of 10 in 85% ethanol with sodium
borohydride followed by peroxidation of the resulting mixture of
homoallyl alcohols 11 with m-CPBA in the presence of NaHCO3
to 12 and, finally, Swern oxidation to give the hydroxyenone 13
in 57% overall yield from 10. Oxidation by the peracid was
assumed (later confirmed) to take place from the α side because
of interference on the β side by the methyl group, which is axial to
the B ring. The next task was to reduce the enone in 13 to the
saturated trans-fused ketone. Chemical reduction, such as Li/
NH3, would be unsuitable because of the elimination of the oxide
bridge by any process that would add electrons to the enone
system. Catalytic hydrogenation was required, and we hoped that
the desired stereochemistry would be favored by putting a bulky
TBS group on the axial 7α-hydroxyl, as shown in 14. In fact,
reduction of 14 with 10% Pd on charcoal, in the presence of
sodium acetate, gave a saturated ketone that, we assumed, could
be represented by 15. The correctness of the various assumptions
implied in structure 15 was verified by an X-ray structure
determination.
We now faced the problem of converting the cyclohexanone
ring of 15 into a cyclopentane system (the C ring), functionalized
to allow the construction to proceed by transforming 15 into the
cyclopentene aldehyde 20. Conversion of 15 to 20 was started by
the formation of the TMS enol ether (LHMDS, followed by
TMS-Cl-Et3N), which we expected would be mostly the single
isomer 16 because of the preference for a double bond parallel to
the ring junction in a trans-decalin system.14 Treatment with cat.
OsO4 for 2 h at rt and reduction of the ketol 17 (NaBH4,
The dienophile 23 was synthesized by a Wadsworth−Emmons
condensation between siloxyaldehyde 2417 and the phosphonate
25, accessible by the coupling of 2-cyano-5-methylpyridine18
with dimethyl methylphosphonate. The Diels−Alder adduct 26
B
DOI: 10.1021/acs.orglett.7b02434
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
was then obtained, in refluxing toluene with 23, in the presence
of the inhibitor, 3-tert-butyl-4-hydroxy-5-methylphenyl sulfide,
in about 88% yield, apparently as a single substance (Scheme 4).
with silica gel that removed the primary TMS), to give the
alcohol 29. Cyclization, followed by treatment with triflic
anhydride and 2,6-di-tert-butylpyridine in DCM, 1 h at −78
°C, led to the pyridinium salt 30, which was immediately
hydrogenated, using platinum oxide in MeOH in the presence of
potassium acetate. We expected the hydrogenation to take place
from the α side to give 31 because of the hindrance by the C20
axial tertiary TMS ether on the β side. Survival of the olefinic
bond in the ring D was, presumably, the result of hindrance by
the axial benzyl ether group on one side, and the C7 axial TBS
ether group on the other. All of these anticipations were satisfied
by the X-ray structure determination of 31.
Scheme 4. Diels−Alder Cycloaddition, Grignard Reaction,
and Construction of E−F Ring (Completion of A−B−C−D−
E−F)
Before OsO4 oxidation, which was to follow, removal of both
silyl groups, the TMS group (E ring) and the TBS group (B ring),
seemed desirable. But, after many attempts, removal of the
tertiary TMS group was unsuccessful, presumably because of the
high steric hindrance in this particular case. We had partial
success with cesium fluoride (CsF); a large excess of CsF and 31,
in anhyd DMF solution, were heated at 95−100 °C for 70 h. At
this point, 57% of 32a in which both the TMS on the C20
hydroxyl and the TBS group on the C7 hydroxyl had been
removed, was yielded along with 32b in which only the TBS
group was removed. Compound 32b was converted to 32a, by
applying the same reaction conditions (CsF, 95−100 °C, 70 h).
We were conscious that the osmylation could add hydroxyls
either from the α side, as desired, or from the β side (Scheme 5).
Scheme 5. Introduction of Diol and Deprotection
We were confident that the correct stereoisomer had resulted
because (1) the dienophile would approach the more accessible
face of the diene, i.e., cis to the starred hydrogen (H*) in 22, and
(2) there should be exclusive endo addition because exo addition
would lead to very severe interference between the TBSO
substituent on the dienophile and the diene. As a result of such
endo addition, the benzyloxy group in the diene and the
acylpyridine of the dienophile would end up cis to each other in
the adduct.
The adduct 26 is capable of easy elimination of the benzyloxy
group, so that risk must be reduced by making the next step, the
addition of the methyl Grignard reagent at low temperature to
the keto group of 26. This could give either of two stereoisomers
or a mixture of the two. After many experiments, it was
concluded that the major product from MeMgBr with 26 had the
undesired stereochemistry at the newly formed tertiary alcohol
center.19 While this was disappointing, it had been shown by H.
Yamamoto20 that the stereochemistry of Grignard reactions with
cyclic ketones could be reversed by previous complexation with
the trimethylaluminum salt of 2,6-di-tert-butyl-4-methylphenol
(BHT). It was gratifying that this process (toluene, −78 °C)
gave, in 89% yield, a tertiary methyl carbinol different from the
one previously obtained and clearly the desired tertiary alcohol
27a, in addition to 11% of 27b, the undesired isomer obtained
previously.
Some assistance to the α side hydroxylation might come from the
7-hydroxyl in ring B, and, additionally, there was interference to
coming from the β side because of the benzyloxy group emerging
from that side. Anyway, we were ready to try the osmylation: 32a
was first protonated with a slight excess of PPTS in dry pyridine,
followed by stirring for 20 h at rt under argon, with about a 20%
excess of OsO4, to give a high yield of two glycols (correct MW)
in a 92:8 (33a:33b) ratio. The major isomer 33a was obtained as
solid, but, unfortunately, could not be crystallized. Our
expectation that 33a was the desired glycol was confirmed by
NOE interproton distance calculations, combined with DFT
calculations (QM/NMR).21 The distance between the hydro-
gens at C15 and C7 in the DFT-optimized geometry of
diastereomer 33a (desired glycol) was measured as 2.58 Å, as it is
in the X-ray of germine (1),1c while the distance between the
same hydrogens in the undesirable glycol 33b was measured as
3.68 Å. The interproton distance between C15 and C7 calculated
from the NOE effect in our synthetic glycol (33a) was 2.58 Å,
matching to that for the desired glycol 33a (the details of the
study on the stereochemistry of the two newly created
We now undertook the reduction of the pyridine ring. First, it
was necessary to protect the tertiary alcohol of the diol 28 to
prevent the participation of the tertiary hydroxyl in reactions
such as cyclization to a tetrahydrofuran ring. A TMS ether was
selected for that role (excess TMS-Cl, followed by treatment
C
DOI: 10.1021/acs.orglett.7b02434
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
the synthesis of the ( )-germine derivative rather than the (+)-isomer,
reducing the relative asymmetric center from 15 to 14.
(4) Stork, G.; Meisels, A.; Davies, J. E. J. Am. Chem. Soc. 1963, 85, 3419.
(5) Kabalka, G. W.; Yang, D. T. C.; Baker, J. D. J. Org. Chem. 1976, 41,
574.
(6) Compound 4 was previously observed in the mixture of products
from the degradation of abietc acid: Wirthlin, T.; Wehrli, H.; Jeger, O.
Helv. Chim. Acta 1974, 57, 351.
(7) Epoxidation from the α-face of the olefin in 4 is favored due to a
steric hindrance from the angular methyl group at the C10, blocking the
β-face (more than a steric hindrance from the C6 methylene group on
the α-face of the olefin).
(8) Treatment with either lithium diethylamide in THF or aluminum
tert-butoxide in refluxing toluene, undesired α-expoxide 5α can be
opened to give the undesired α-allylic alcohol (an epimer of 6). Further
treatment of the undesired α-allylic alcohol with Al(iPrO)3 (m-xylene,
135 °C) equilibrates to the desired more stable β-allylic alcohol 6 (1:4 =
α/β). Further heating did not alter the ratio. Thus, a mixture of 5α and
5β was treated directly with Al(iPrO)3 in boiling m-xylene. (a) Dauben,
W. G.; Fonken, G. J.; Noyce, D. S. J. Am. Chem. Soc. 1956, 78, 2579.
(b) Eliel, E. L.; Ro, R. S. J. Am. Chem. Soc. 1957, 79, 5992.
(9) Van Rheenen, V.; Kelly, R. C.; Cha, D. Y. Tetrahedron Lett. 1976,
17, 1973.
(10) We predicted that OsO4 oxidation might be favored to proceed
from the α-side of the olefin because the β-face of the olefin is severely
hindered by an angular (axial) methyl group and the cis A/B ring
junction. This suggestion was supported by the success of the
subsequent aromatic ring oxidation, a reaction possible only when the
newly formed tertiary hydroxy group is both α and axial on the A ring.
stereocenters at C14 and C15 are described in the SI). Thus, the
minor component 33b was determined to be the undesired
glycol.
Now only one step remained, removal of the benzyl protecting
group from the C16 hydroxyl. Because of the well-known
poisoning effect of free amino groups on palladium hydro-
genation, the amino group in 34a was first covered by making its
hydrochloride, by treatment with hydrogen chloride in MeOH
(the acetonide was also eliminated by that operation). This was
followed by hydrogen, pressurized to 700 psi, for 20 h at rt with
10% Pd on charcoal. The yield of debenzylation was essentially
quantitative, giving the structure 34 (the hydrogen chloride salt
of 2, Figure 1).
In conclusion, compound 34 is racemic germine hydro-
chloride having an extra methylene group at C4 but with all of 15
of germine’s asymmetric centers3 correctly established relative to
the single asymmetric center of the starting material 3.22
ASSOCIATED CONTENT
* Supporting Information
■
S
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.orglett.7b02434.
X-ray structures for 5, 6, 15, 21, 31 and 35 (ZIP)
1
Experimental procedures and characterization data, H-
and 13C NMR spectra, LCMS data; study on the
stereochemistry of the C14-OH and the C15-OH in 33a
using QM calculation and NOE effects; NMR−NOE
spectra of 33a in MeOH-d4 and benzene-d6 (PDF)
(11) (a) Hata, K.; Hamamoto, H.; Shiozaki, Y.; Cammerer, S. B.; Kita,
̈
Y. Tetrahedron 2007, 63, 4052. (b) Danishefsky, S. J.; Dai, M.
Heterocycles 2009, 77, 157.
(12) This process was well known in the case of steroid dienones:
Djerassi, C.; Gutzwiller, J. J. Am. Chem. Soc. 1966, 88, 4537.
(13) Marshall, J. A.; Greene, A. E. J. Org. Chem. 1971, 36, 2035.
(14) House, H. O.; Trost, B. M. J. Org. Chem. 1965, 30, 1341. Mazur,
Y.; Sondheimer, F. J. Am. Chem. Soc. 1958, 80, 5220.
(15) We also made 21 by a different route: sodium borohydride, then
benzoyl chloride on the aldehyde 20, in which the ring A glycol had been
deprotected and transformed into a mono-p-methoxybenzyl ether. See
the SI for the conversion of 20 to 21.
AUTHOR INFORMATION
■
Corresponding Author
*E-mail: gjs8@columbia.edu.
ORCID
Ayako Yamashita: 0000-0002-7601-9969
Keith Jones: 0000-0002-9440-4094
(16) (a) Mandai, T.; Osaka, K.; Wada, T.; Kawada, M.; Otera, J.
Tetrahedron Lett. 1983, 24, 1171 (E/Z = 2.5/1). (b) Mandai, T.; Osaka,
K.; Kawagishi, M.; Kawada, M.; Otera, J. J. Org. Chem. 1984, 49, 3595.
This paper is perhaps the most relevant: it makes conjugated
alkoxydienes via triphenylphosphoranes and obtains the dienes in an
E/Z ratio of 2.5:1. (c) Ikeda, S.; Shibuya, M.; Kanoh, N.; Iwabuchi, Y.
Chem. Lett. 2008, 37, 962 (E/Z = 5/2). (d) Patel, P. R.; Boger, D. L. Org.
Lett. 2010, 12, 3540 (E/Z = 1/1.5). (e) Earnshaw, C.; Wallis, C. J.;
Warren, S. J. Chem. Soc., Perkin Trans. 1 1979, 3099.
(17) (a) McDougal, P. G.; Rico, J. G.; Oh, Y.-I.; Condon, B. J. Org.
Chem. 1986, 51, 3388. (b) Aszodi, J.; Bonnet, A.; Teutsch, G.
Tetrahedron 1990, 46, 1579. We made this compound, originally, by
ozonolysis of the bis TBS derivative of cis 1,4-butenediol.
(18) Crabb, T. A.; Heywood, G. C. Org. Magn. Reson. 1982, 20, 242.
(19) The stereochemistry of 27b was confirmed by X-ray analysis of
the derivative 35. See the SI for the conversion of 27b to 35.
(20) Maruoka, K.; Itoh, T.; Sakurai, M.; Nonoshita, K.; Yamamoto, H.
J. Am. Chem. Soc. 1988, 110, 3588.
(21) Chini, M. G.; Jones, C. R.; Zampella, A.; D’Auria, M. V.; Renga, B.;
Fiorucci, S.; Butts, C. P.; Bifulco, G. J. Org. Chem. 2012, 77, 1489.
(22) A plan for conversion of 33a to 1 was (with various
deprotections/protections) C4-CH2OH to C4-CO2H, followed by
Barton’s conditions to change C4-CO2H to C4-OH. Barton, D. H. R.;
Gero, S. D.; Holliday, P.; Quiclet-Sire, B.; Zard, S. Z. Tetrahedron 1998,
54, 6751. At this point, we realized that we did not have enough material
(a few milligrams) to go through the several steps for this conversion.
One would have to restart the whole synthesis. But I (G.S.) am now 95
years old...
Masato Kitamura: 0000-0002-0543-8795
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
This research was supported by the NIH (R01 GM05147, R01
HL25635) and NSF (CHE-86-12434). We acknowledge our
colleagues Professor Gerard Parkin for the X-ray analyses,
Professor James Leighton for helping with the high-pressure
hydrogenation, Professor W. Clark Still for helping distinguish-
ing 33a,b, and Dr. John Decatur for NOE experiments for 33a.
Dr. Anil K. Saksena of Schering-Plough Corp. is thanked for a
generous supply of germine.
REFERENCES
■
(1) (a) Greenhill, J. V.; Graysham, P. T. Alkaloids; Academic Press,
1992; Vol. 41, pp 177−187. (b) Kupchan, S. M.; Narayanan, C. R. J. Am.
Chem. Soc. 1959, 81, 1913. (c) Tang, J.; Li, H.-L.; Shen, Y.-H.; Jin, H.-Z.;
Yan, S.-K.; Liu, R.-H.; Zhang, W.-D. Helv. Chim. Acta 2007, 90, 769
(CCDC-635769).
(2) (a) Baker, P. D. Southern Med. Surg. 1859, 15, 579. (b) Renforado,
J. M.; Flacke, W.; Swaine, C. R.; Mosimann, W. J. Pharmacol. Exp. Ther.
1960, 130, 311. (c) Saksena, A. K.; McPhail, A. T. Tetrahedron Lett.
1982, 23, 811.
(3) The equilibrium between hemiketal ⇌ hydroxy ketone reduced the
number of asymmetric centers to 15 in (+)-germine. Our interest was in
D
DOI: 10.1021/acs.orglett.7b02434
Org. Lett. XXXX, XXX, XXX−XXX