Catalytic oxidations of steroid substrates by artificial cytochrome p-450 enzymes.

Article abstract of DOI:10.1021/jo020174u

Catalysts comprising manganese-porphyrins carrying cyclodextrin binding groups are able to perform hydroxylations with substrate selectivity and regio- and stereoselectivity and high catalytic turnovers. The geometries of the catalyst/substrate complexes override intrinsic substrate reactivities, permitting attack on geometrically accessible saturated carbons of steroids in the presence of secondary carbinol groups and carbon-carbon double bonds, as in enzymatic reactions. Selective hydroxylations of steroid carbon 9 positions are of particular practical interest.

Full text of DOI:10.1021/jo020174u

VOLUME 67, NUMBER 15
J ULY 26, 2002
© Copyright 2002 by the American Chemical Society
Ca ta lytic Oxid a tion s of Ster oid Su bstr a tes by Ar tificia l
Cytoch r om e P -450 En zym es†
J erry Yang,‡ Bartolo Gabriele,§ Sandro Belvedere, Ying Huang, and Ronald Breslow*
Department of Chemistry, Columbia University, New York, New York 10027
rb33@columbia.edu
Received March 12, 2002
Catalysts comprising manganese-porphyrins carrying cyclodextrin binding groups are able to
perform hydroxylations with substrate selectivity and regio- and stereoselectivity and high catalytic
turnovers. The geometries of the catalyst/substrate complexes override intrinsic substrate reac-
tivities, permitting attack on geometrically accessible saturated carbons of steroids in the presence
of secondary carbinol groups and carbon-carbon double bonds, as in enzymatic reactions. Selective
hydroxylations of steroid carbon 9 positions are of particular practical interest.
In tr od u ction
selectivity in enzymatic reactions is so common, and so
unlike normal organic synthetic selectivity, that fermen-
tation methods are often used as part of pharmaceutical
manufacturing. For example, biological fermentations are
used in the synthesis of corticosteroids to achieve the
oxygenation of ring C in steroids, replacing much more
tedious chemical methods to produce the compounds.2b,c
Regioselectivity in organic synthesis is normally
achieved by the manipulation of substrate functional
groups. Reactions occur at the position of the substrate
that has the greatest intrinsic reactivity (sometimes
modified by neighboring groups, as in the selective
reactivity of the double bonds of allylic alcohols toward
coordinating reagents1a or the intramolecular attack of
a substrate functional group on another position.1b) If this
is not desired, that most reactive position can be blocked
with a protecting group or another position activated by
changing the functional group. That is, regioselectivity
in chemical synthesis is typically dominated by the
competitive reactivity of sections of the substrate toward
relatively random reagents.
We have been developing chemical methods that
imitate such geometric control, to achieve the kinds of
regioselectivities heretofore possible only with enzymes.
We coined the term biomimetic chemistry to describe
such methods.3 In the earliest approach, we showed that
a benzophenone ester of a steroid could undergo intra-
molecular photochemical hydrogen abstractions, afford-
ing regioselective functionalization of the steroid remote
from the ester attachment point.4 We referred to this
process as remote functionalization.5 Benzophenone es-
ters with different geometries afforded different products.
By contrast, regioselectivity in biochemistry is achieved
as the result of the geometry of an enzyme/substrate
complex, and that regioselectivity can completely override
the normal reactivity pattern of the substrate. A striking
example is the selective oxidation of three unactivated
methyl groups in lanosterol (1) by a cytochrome P-450
enzyme in the biosynthesis of cholesterol (2), while two
double bonds and a secondary carbinol group are left
untouched.2a Such dominant geometric control of regio-
* To whom correspondence should be addressed. Tel: 212-854-2170.
Fax: 212-854-2755.
† Taken in part from the Ph.D. theses of J erry Yang, Sandro
Belvedere, and Ying Huang.
(2) (a) Steroid Hormone Synthesis. In Encyclopedia of Human
Biology; Academic Press: San Francisco, 1991; Vol. 7. (b) Dawson, J .
H.; Sono, M. Chem. Rev. 1987, 87, 1255. (c) Woggon, W. D. Top. Curr.
Chem. 1996, 184, 39.
‡ Present address: Harvard University, Department of Chemistry
and Chemical Biology, 12 Oxford Street, Cambridge, MA 02138.
§
Present address: Dipartimento di Scienze Farmaceutiche,
Universita’della Calabria, 87036 Arcavacata di Rende, Cosenza, Italy.
(1) (a) Sharpless, K. B.; Michaelson, R. C. J . Am. Chem. Soc. 1973,
95, 6136. (b) E.g., the Barton reaction: Barton, D. H. R.; Beaton, J .;
Geller, L.; Pechet, M. J . Am. Chem. Soc. 1961, 83, 4076.
(3) Breslow, R. Chem. Soc. Rev. 1972, 1, 553.
(4) Breslow, R.; Baldwin, S.; Flechtner, T.; Kalicky, P.; Liu, S.;
Washburn, W. J . Am. Chem. Soc. 1973, 95, 3251.
(5) Breslow, R.; Winnik, M. A. J . Am. Chem. Soc. 1969, 91, 3083.
10.1021/jo020174u CCC: $22.00 © 2002 American Chemical Society
Published on Web 06/06/2002
J . Org. Chem. 2002, 67, 5057-5067
5057

Yang et al.
We then developed methods to achieve remote func-
tionalization by free radical chain reactions. In the first
method, we attached a phenyliodine dichloride species
to the steroid, again as an ester, and saw that upon
irradiation it chlorinated specific carbons remote from
the ester attachment point.6 In a later related scheme,
we attached only an iodophenyl group to the substrate
as a template and saw that it directed selective chlorina-
tion by a process we described as a radical relay reaction.7
It was also a catalyst for the reactionswithout the iodine
atom there was no substrate chlorination under the
reaction conditions. Other templates and radical reac-
tions were developed as well and their geometries
directed the halogenations to other selected spots on the
substrates.8
These methods are powerful, and indeed, using the
radical relay method, we achieved the synthesis of
cortisone, in which the ring C functionalization was
geometrically directed, as in an enzyme.9 However, there
were drawbacks to these methods. The reagent or tem-
plate was attached, or in one case coordinated,10 to the
substrate by a single linkage, so there was considerable
flexibility. The geometric control was chiefly a control of
the distance of the attack point from the attachment
point. Also, the reagents had some intrinsic chemical
selectivity, and functionalizations were partly directed
by the greater reactivity of tertiary hydrogens relative
to secondary or primary hydrogens.
To achieve completely geometric control of regioselec-
tivity, it would be necessary to have two or even three
well-defined interactions between substrate and catalyst,
but this becomes unreasonable with covalent attach-
ments. The template catalysts in remote halogenation
can be recovered after the process, but there is not true
turnover catalysis by such a process (although in one case
we attached three substrates to such a template and saw
that all were selectively halogenated, with thus a turn-
over of 3).11 To justify developing a catalyst that could
hold a substrate with more than one binding interaction,
we need true turnover catalysis so that an expensive
catalyst, in terms of the chemical effort needed to make
it, can process hundreds or thousands of substrate
molecules. Thus, the catalyst/substrate interactions should
not use stable covalent bonds.
In one approach, we examined a system in which two
ion-pairing interactions stretched a flexible substrate
above a benzophenone unit.12 Photolysis did indeed steer
reaction into a geometrically well-defined spot, but this
is of course not a catalytic reaction. For turnover cataly-
sis, we made a catalyst with an iron porphyrin unit
carrying four hydroxyquinoline metal-binding groups. We
saw that its Cu2+ complex could bind a substrate having
two metal coordinating groups and catalyze the epoxi-
dation of its double bond with eight turnovers and good
selectivity over noncoordinating substrates.13 We then
turned to hydrophobic binding into cyclodextrins in water
solution.
The metalloporphyrin system has been developed
extensively to mimic cytochrome P-450 enzymes.14 We
attached cyclodextrin units to it so it could reversibly bind
substrates in well-defined positions. The aim was to
achieve selective oxidations of the bound substrates
directed entirely by the geometry imposed by multipoint
binding. Performing the reactions in water also has
environmental advantages, since water is an environ-
mentally benign solvent that can be easily purified. As
expected, we saw that performing the reactions in other
solvents greatly diminished or erased the yields of
products, supporting the importance of hydrophobic
binding.
Two related studies on biomimetic regioselectivity
should be mentioned here. Groves and Neumann exam-
ined the epoxidation and hydroxylation of steroids in a
lipid bilayer with a metalloporphyrin carrying hydropho-
bic side chains.15 They observed good selectivity, but
strong product binding to the catalyst prevented catalytic
turnover. Also, Grieco attached a manganese porphyrin
to a steroid covalently and saw that the system could
perform directed steroid hydroxylation controlled by the
distance between the attachment point and the metal oxo
species.16 This covalent system is related to the ben-
zophenone and iodophenyl systems described above, and
like them it cannot produce turnover catalysis.
Some of our work has been described in preliminary
form elsewhere.17-20 In this paper, we describe the details
of part of the published work and the extension to
important new areas.
Resu lts a n d Discu ssion
â-Cyclodextrin (â-CD) is a cyclic heptaglucose oligomer
that dissolves in water and binds hydrophobic species
into its central space. Using the natural cytochrome
P-450 enzyme as our model, we synthesized some met-
alloporphyrin catalysts that retain the natural porphyrin
catalytic core and have attached â-CDs to produce an
artificial binding pocket (Figure 1). The first example, 3,
with two cyclodextrins and an iron porphyrin core, was
poorly water soluble, and the catalyst itself was easily
(13) Breslow, R.; Brown, A. B.; McCullough, R. D.; White, P. W. J .
Am. Chem. Soc. 1989, 111, 4517-4518.
(14) Meunier, B., Ed. Biomimetic Oxidations Catalyzed by Transition
Metal Complexes; Imperial College Press: London, 2000.
(15) (a) Groves, J . T.; Neumann, R. J . Am. Chem. Soc. 1987, 109,
5045-5047. (b) Groves, J . T.; Neumann, R. J . Org. Chem. 1988, 53,
3891-3893. (c) Groves, J . T.; Neumann, R. J . Am. Chem. Soc. 1989,
111, 2900-2909.
(6) Breslow, R.; Dale, J . A.; Kalicky, P.; Liu, S. Y.; Washburn, W.
N. J . Am. Chem. Soc. 1972, 94, 3276.
(7) Breslow, R.; Corcoran, R. J .; Snider, B. B. J . Am. Chem. Soc.
1974, 96, 6791.
(8) (a) Breslow, R.; Wife, R. L.; Prezant, D. Tetrahedron Lett. 1976,
1925. (b) Breslow, R.; Brandl, M.; Hunger, J .; Adams, A. D. J . Am.
Chem. Soc. 1987, 109, 3799-3801. (c) Breslow, R.; Guo, T. Tetrahedron
Lett. 1987, 28, 3187-3188. (d) Batra, R.; Breslow, R. Tetrahedron Lett.
1989, 30, 535-538. (e) Breslow, R.; Wiedenfeld, D. Tetrahedron Lett.
1993, 34, 1107-1110.
(16) (a) Grieco, P. A.; Stuk, T. L. J . Am. Chem. Soc. 1990, 112, 7799.
(b) Kaufman, M. D.; Grieco, P. A.; Bougie, D. W. J . Am. Chem. Soc.
1993, 115, 11648. (c) Stuk, T. L.; Grieco, P. A.; Marsh, M. M. J . Org.
Chem. 1991, 56, 2957.
(17) , Breslow, R.; Zhang, X.; Xu, R.; Maletic, M.; Merger, R. J . Am.
Chem. Soc. 1996, 118, 11678-11679.
(9) Breslow, R.; Snider, B. B.; Corcoran, R. J . J . Am. Chem. Soc.
1974, 96, 6792.
(18) Breslow, R.; Huang, Y.; Zhang, X.; Yang, J . Proc. Natl. Acad.
Sci. U.S.A. 1997, 94, 11156-11158.
(10) Breslow, R.; Heyer, D. Tetrahedron Lett. 1983, 24, 5039.
(11) Breslow, R.; Heyer, D. J . Am. Chem. Soc. 1982, 104, 2045.
(12) Breslow, R.; Rajagopalan, R.; Schwarz, J . J . Am. Chem. Soc.
1981, 103, 2905.
(19) , Breslow, R.; Gabriele, B.; Yang, J . Tetrahedron Lett. 1998,
39, 2887-2890.
(20) Yang, J .; Breslow, R. Angew. Chem., Int. Ed. 2000, 39, 2692-
2694.
5058 J . Org. Chem., Vol. 67, No. 15, 2002

Catalytic Oxidations of Steroid Substrates
F IGURE 1. Early catalyst 3 (FeIII form) and the better catalyst 4 (MnIII form). Substrate 5 binds to 4 and is hydroxylated with
hundreds of turnovers.
destroyed oxidatively. Thus, we turned to catalyst 4,
carrying four cyclodextrin rings to increase water solubil-
ity. As metal we used manganese, not the iron used by
the enzyme; the iron analogue of 4 was also effective in
our system, but less so than the manganese version.
Groves has reported such a catalytic preference for
manganese in cytochrome P-450 model systems.21 Cata-
lyst 4 was able to bind a simple dihydrostilbene substrate
5 in water and catalyze the hydroxylation of saturated
carbons, using excess iodosobenzene as oxidant. There
were up to 650 turnovers and rates somewhat faster than
those of natural enzymes for similar benzylic oxidations.18
Oxidants other than iodosobenzene have been used in
such P-450 model systems,22 but we find that iodosoben-
zene is the best reagent in this and the work to be
described subsequently. Also, it was important to steer
the oxygen atom bonded to the metal, and the bound
substrate, to the same face of the catalyst. Thus, in this
and the remaining work to be described we added several
equivalents of pyridine to bind to one face of the metal-
loporphyrin and steer substrate and oxygen to the other
face; pivalate ion and imidazole were also able to do this,
but less effectively than pyridine. At the end of this paper,
we will mention a new catalyst in which the pyridine ring
is part of the catalyst, and no external pyridine is
required.
Such hydroxylations involve the formation of an in-
termediate porphyrin metal oxo compound, which then
oxidizes a C-H bond. There has been some dispute about
the hydroxylation mechanism. One choice, proposed by
Groves, is the oxygen rebound mechanismsthe metal oxo
species abstracts hydrogen from the substrate to produce
a substrate radical, which then captures the hydroxyl
group from the metal.23 The other choice is a direct
oxygen insertion into the C-H bond by the metal oxo
species.24 A theoretical treatment apparently supports
the oxygen rebound mechanism.25
We examined the epoxidation of some stilbene deriva-
tives that could bind into catalyst 4 and saw sensible
substrate selectivity.17 In line with our modeling, we
found that an analogue of 4 carrying only two trans
(23) (a) Groves, J . T.; McClusky, G. A.; White, R. E.; Coon, M. J .
Biochem. Biophys. Res. Commun. 1978, 81, 154. (b) Groves, J . T.;
Subramanian, D. V. J . Am. Chem. Soc. 1984, 106, 2177.
(24) (a) Newcomb, M.; Shen, R.; Choi, S. Y.; Toy, P. H.; Hollenberg,
P. F.; Vaz, A. D. N.; Coon, M. J . J . Am. Chem. Soc. 2000, 122, 2677.
(b) Newcomb, M.; Toy, P. H. Acc. Chem. Res. 2000, 33, 449. (c)
Newcomb, M.; Le Tadic-Biadatti, M. H.; Chestney, D. L.; Roberts, E.
S.; Hollenberg J . Am. Chem. Soc. 1995, 117, 12085. (d) Toy, P. H.;
Newcomb, M.; Coon, M. J .; Vaz, A. D. N. J . Am. Chem. Soc. 1998, 120,
9718.
(21) Groves, J . T.; Kruper, W. J .; Haushalter, R. C. J . Am. Chem.
Soc. 1980, 102, 6375.
(22) (a) Collman, J . P.; Tanaka, H.; Hembre, R. T.; Brauman, J . I.
J . Am. Chem. Soc. 1990, 112, 3689. (b) Nee, M. W.; Bruice, T. C. J .
Am. Chem. Soc. 1982, 104, 6123. (c) Mansuy, D.; Bartoli, J . F.;
Chottard, J . C.; Lange, M. Angew. Chem., Int. Ed. Engl. 1980, 19, 909.
(d) Robert, A.; Meunier, B. New J . Chem. 1988, 12, 885. (e) Traylor, P.
S.; Dolphin, D.; Traylor, T. G. J . Chem. Soc., Chem. Commun. 1984,
279.
(25) (a) Ogliaro, F.; Harris, N.; Cohen, S.; Filatov, M.; de Visser, S.
P.; Shaik, S. J . Am. Chem. Soc. 2000, 122, 8977. (b) Hata, M.; Hoshino,
T.; Tsuda, M. Chem. Commun. 2000, 6383.
J . Org. Chem, Vol. 67, No. 15, 2002 5059

Yang et al.
F IGURE 2. Selective hydroxylation of substrate 6 with catalyst 4, forming only the C-6 alcohol 7, hydrolyzed to triol 8.
cyclodextrins and two unsubstituted phenyls was a
catalyst comparable to 4, but the analogous cis isomer
was not. For catalysis, the substrates do indeed need to
bind across the porphyrin ring.
zene it catalyzed the clean conversion of 6 to 7 in water
with the same selectivity as with 4.19 With 0.1 mol %
catalyst there was 18.7% conversion of 6 to 7, indicating
187 turnovers. With 1 mol % catalyst there was again
100% yield of 7, but with only 95 turnovers, so at high
conversion there might be some inhibition by product
binding to the catalyst. These carefully obtained numbers
came from an analysis involving complete conversion of
the alcohols to benzoate esters, and then calibrated
quantitative hplc analysis of the benzoates.
Fluorination of the phenyl rings has been reported
to stabilize tetraarylporphyrins against autoxidative
destruction.26-28 We created a particularly easy synthesis
of 9sreaction of the porphyrin 10 carrying pentafluo-
rophenyl rings with the cyclodextrinthiol (Figure 3). The
para fluorines are selectively replaced; ortho and para
positions are activated to nucleophilic attack, and the
ortho positions are hindered.
We examined the effect of converting the sulfide links
in 4 to sulfone groups, which might occur during oxida-
tion reactions. We synthesized the corresponding tetra-
sulfone (see the Experimental Section) by oxidation of
the porphyrin derivative with magnesium monoperoxy-
phthalate, followed by manganese incorporation, and
found that it was also a catalyst for the hydroxylation of
5, but with only 18% of the yield seen with 4. Thus, the
sulfide links are better than sulfones in the catalyst.
Of course, there is no unusual positional selectivity in
the oxidation of 5, but we did see that an analogue of 5
lacking the tert-butylphenyl groups was not oxidized
under these conditions, so there was substrate selectivity.
In a more striking and important case, we saw that
catalyst 4 was able to bind a steroid 6 carrying two tert-
butylphenyl binding groups along with water-solubilizing
sulfonate groups, but not an analogue without the tert-
butylphenyl binding group, and perform regioselective
hydroxylation at the steroid C-6 position (Figure 2).18 The
reaction was also stereospecific, yielding only the equato-
rial C-6 alcohol 7, whose hydrolysis afforded triol 8.
However, this reaction proceeded with only four to five
turnovers, in contrast to the 650 turnovers for the more
reactive 5. After these turnovers, UV/vis spectra indi-
cated that the porphyrin system had been oxidatively
destroyed.
The specific hydroxylation of a steroid at C-6 is
interesting, and the finding that the product is a second-
ary alcohol is striking. When steroids are oxidized with
high concentrations of unbound porphyrins and iodo-
sobenzene, we find that steroidal secondary alcohols are
normally simply oxidized to ketones, before any hydroxy-
lations of saturated carbons occur. This reflects the
(26) (a) Traylor, T. G.; Tsuchiya, S. Inorg. Chem. 1987, 26, 1338.
(b) Traylor, P. S.; Dolphin, D.; Traylor, T. G. J . Chem. Soc., Chem.
Commun. 1984, 279.
(27) (a) Wijesekera, T.; Matsumoto, A.; Dolphin, D.; Lexa, D. Angew.
Chem., Int. Ed. Engl. 1990, 29, 1028. (b) Dolphin, D.; Traylor, T. G.;
Xie, L. Acc. Chem. Res. 1997, 30, 251.
To obtain higher turnover catalysis in the steroid case,
we synthesized catalyst 9 and saw that with iodosoben-
(28) Ellis, P. E.; Lyons, J . E. Coord. Chem. Rev. 1990, 105, 181.
5060 J . Org. Chem., Vol. 67, No. 15, 2002

Catalytic Oxidations of Steroid Substrates
F IGURE 3. Synthesis of the fluorinated catalyst 9 from porphyrin 10.
F IGURE 4. Computer model of the complex of substrate 6 with catalyst 9. For clarity, only the two cyclodextrins involved in
binding are shown. The oxygen atom on the manganese is almost in van der Waals contact with the C-6 equatorial hydrogen, but
the C-6 axial hydrogen is inaccessible.
higher oxidative reactivity of a secondary carbinol. The
fact that we saw no ketone in the oxidation of 6 bound
to 4 or 9, even with many turnovers, is an indication that
geometric control has overcome normal functional group
reactivity. We propose that the oxidation of a carbinol to
a ketone by a metal oxo porphyrin intermediate requires
that the oxygen atom remove the carbinol hydrogen of
the C-H group and that this is not accessible in the
J . Org. Chem, Vol. 67, No. 15, 2002 5061

Yang et al.
F IGURE 5. Computer model of the complex of substrate 11 with catalyst 9. For clarity, the cyclodextrin ring not involved in
binding is hidden. The oxygen atom on the manganese is in van der Waals contact with the C-9 hydrogen.
epoxidation and HF opening affords 9-fluoro-11-hydroxy
steroids that are useful corticosteroid derivatives. Thus,
we determined to alter the geometry of our catalyst/
substrate complexes so as to direct hydroxylation to C-9
of the steroid.
It was clear from the result and from molecular and
computer models (see Figure 4) that the C-6 hydroxyla-
tion occurs because the complex of substrate 6 into the
catalysts 4 and 9 prefers a geometry in which the steroid
edge projects toward the porphyrin unit. To achieve
hydroxylation at C-9, it seemed that we simply had to
change this geometry so that the steroid R face was above
the MndO species. Models showed (see Figure 5) that
this would occur if we made a triple ester of product triol
8 so that three of the cyclodextrins in catalyst 9 could
bind the substrate. This is what occurred.
Triol 8 was converted to a triple binding steroid
substrate 11. When this was oxidized by catalyst 9, the
C-9-hydroxylated product 12 was indeed found to be the
only significant product of the reaction (Figure 6),
consistent with the models. Ester hydrolysis converted
the hydroxytriester to the tetraol 13. With 1 mol % 9,
substrate 11 was converted to 13 with 72% conversion
and 100% yield, indicating 72 turnovers. With 0.1 mol %
9, product 13 was formed in 8% conversion but again
100% yield, so there were 80 turnovers. We never saw
any C-9 hydroxylation of the double binding substrate 6
or of its 6-hydroxylated product 7, so the geometry forced
by the triple binding of 11 is apparently not accessed
without the extra binding group.
F IGURE 6. Conversion of triol 8 to substrate 11 and its
hydroxylation at C-9 to form product 12, which was hydrolyzed
to tetraol 13.
complex of product 7 with the catalysts. The â-hydrogen
on C-6 is axial, and models (see Figure 4) show that it is
not accessible in the product/catalyst complex.
We also synthesized a perhalogenated version 14 of our
catalyst, differing from 9 by the incorporation of chlorine
atoms at the â-pyrrole porphyrin positions. Catalyst 14
showed a 2-fold increase in efficiency compared with 9,
oxidizing 6 at C-6 with up to 356 turnovers and 11 at
C-9 with up to 152 turnovers.
Hydroxylation of a steroid at C-9 or at C-11 would be
more important. Both these positions can be hydroxylated
by enzymes in fermentation processes2 and the resulting
carbinols dehydrated to afford a 9(11) double bond. Then
5062 J . Org. Chem., Vol. 67, No. 15, 2002

Catalytic Oxidations of Steroid Substrates
to position steroid carbons in ring D over the reactive
metal-oxo species in the catalyst-substrate complex.
In our first rationally designed substrate, we removed
the phenyl ring on the linker attached at the steroid 3
position of substrate 6 and inserted it on the phenyl ring
of the linker attached at the 17 position of the steroid.
The resulting substrate 17, with a tert-butyl binding
group on the linker at the 3 position of the steroid and a
tert-butylbiphenyl binding group on the linker attached
to the 17 position of the steroid, would change the
position of the steroid in our substrate-catalyst system
without changing the length of the substrate. Further-
more, the steroid should be moved by four carbon atoms
relative to the porphyrin metal center, so oxidation
should occur on the D ring, specifically on C-15, instead
of the original C-6 position.
In recent work,29 we have synthesized a manganese
porphyrin derivative 15 carrying three cyclodextrin
groups and one o-nitrophenyl group. Also, compound 16
was derived from it, with a pyridine ring that can
coordinate to the manganese atom. The hydroxylation of
6 to 7 again occurred with these catalysts, but with 2000
turnovers for 16 and 3000 turnovers for 15. The electron-
withdrawing group of 15 probably diminishes the rate
at which the catalyst itself is oxidatively destroyed, and
there may also be steric effects.
Substrate 17 was subjected to our standard oxidizing
conditions with catalyst 9 at 0 °C, resulting in a mixture
of three steroids after ester hydrolysis. We did see
hydroxylation at C-15, but the 7â-hydroxylated product
18 was produced in a 3:1:1 ratio over the 6R-hydroxylated
8 and the 15R-hydroxylated 19 products, as determined
by HPLC assay. Our prediction that the major hydroxy-
lated product would be at the steroid 15 position was not
correct. This could be due to the ability of the tert-
butylbiphenyl group in 17 to move within the cyclodextrin
cavity such as to bind either phenyl ring in the linker
instead of only the desired ring carrying to the tert-butyl
group.
Therefore, substrate 20, containing two methyl groups
on the tert-butylbiphenyl binding group, was synthesized
so as to prevent this flexibility of movement of the binding
group within the cyclodextrin cavity (Figure 7).
When substrate 20 was subjected to our hydroxylation
conditions, the result was a dramatic suppression of the
7-hydroxylated product 18. However, a 1:1 mixture of the
6R (8) and 15R (19) product was produced. We rationalize
this result by concluding that the tert-butyl binding group
attached at the steroid 3 position in 20 is not hydrophobic
enough to produce tight binding.
Substrate 21 was then synthesized containing an
adamantyl binding group linked to the 3 position of the
steroid. The adamantylglutaric acid group is a strong
binder to â-cyclodextrin, with a measured binding con-
stant (Ka) of approximately 25 000 M-1 in pH 10 phos-
phate buffer. For comparison, 3-tert-butylglutaric acid
was measured to have a Ka of ∼400 M-1 to â-cyclodextrin
under the same conditions. Hydroxylation of substrate
19 resulted in a 2-fold selectivity for the predicted 15R-
hydroxylated product 19 over triols 8 and 18. Thus, we
were indeed able to move the preferred position of
hydroxylation to the predicted C-15 position to some
In our earlier studies on template-directed free-radical
reactions of steroids, we saw that we could change the
position of chlorination by changing the length of the
template group. Now we wanted to see whether we could
change the position of hydroxylation by using double
binding of androstandiol, but with different binding
groups at C-3 and C-17. Our first approach was to attach
hydrophobic binding groups with short linkers to the C3-
OH in 3â,17â-androstanediol and a hydrophobic binding
group with a longer linker on the C17-OH. We expected
that the relative position of the steroid backbone with
respect to the metalloporphyrin would change such as
(29) Breslow, R.; Yang, J .; Yan, J . Tetrahedron 2002, 58, 653-659.
J . Org. Chem, Vol. 67, No. 15, 2002 5063

Yang et al.
to be functionalized, producing some of the 3 ketone along
with 1,3-3,6- and 3,15-diols (R and â at C-15) (Figure
8). The striking finding was that only saturated carbons
were hydroxylated in lithocholenic acid. The binding
geometry of this compound34,35 puts the double bond
inside a cyclodextrin, out of reach of the catalytic group.
Thus, in this case, geometric control in the complex lets
us oxidize saturated carbons without attacking the
normally more reactive double bond or an allylic position.
This finding also indicates that we have enabled geo-
metric control to override intrinsic substrate reactivities,
as in natural enzymes.
Con clu sion s
1. Metalloporphyrins with attached cyclodextrin groups
can bind various steroid derivatives and catalyze their
selective hydroxylations.
2. The selectivities observed are consistent with mo-
lecular and computer models of the complexes.
3. In the best cases, we achieve hundreds and even
thousands of catalytic turnovers of the selective reactions.
4. The geometries of the complexes override intrinsic
reactivities, preventing oxidation of an otherwise reactive
secondary carbinol or of a carbon-carbon double bond.
This mimics selectivity effects typically seen only in
enzymatic reactions.
F IGURE 7. Substrates 20 and 21 move the hydroxylation
position by catalyst 9 to the right, forming (after hydrolysis)
the 7â derivative 18 and the 15R derivative 19, with a
diminished amount of 8.
5. The catalytic selective hydroxylation at carbon 9 in
an androstane derivative provides entry into potential
corticosteroid precursors.
Exp er im en ta l Section
The molecular models shown in Figures 4 and 5 were
obtained from Monte Carlo calculations performed using the
program Macromodel version 7.0 (Schro¨dinger, Inc.), using
water as continuous solvent and energy minimization as the
geometries were varied. Further details of the calculations are
described in the references.20
Syn th eses of th e Ca ta lysts. 4-Methylsulfoxybenzaldehyde
(24) (7.714 g, 46 mmol) was dissolved in 200 mL of propionic
acid under air. The solution was heated to 80 °C and allowed
to stir for 10 min. Pyrrole (3.18 mL, 46 mmol) was added
quickly to the flask, and the solution was heated to reflux at
150 °C for approximately 1 h. The propionic acid was then
removed under reduced pressure, and the solid was run
through a plug of SiO2 with 95/5 CH2Cl2/MeOH. The crude
product was washed several times with pure methanol to
remove the white and blue fluorescent impurities. A total of
four columns (95/5 CH2Cl2/MeOH on SiO2) were used to purify
2 g of the red/purple product 25 (20% yield): 1H NMR (300
MHz, CDCl3) δ 3.06 (s, 12H), 8.05 (d, 8H, J ) 8.2 Hz), 8.37 (d,
8H, J ) 8.2 Hz), 8.84 ppm (s, 8H); 13C NMR (75 MHz, CDCl3/
CD3OD) δ 43.14, 102.87, 118.66, 122.02, 135.06, 144.21, 144.82.
UV/vis (CH2Cl2) λ 418, 513, 548, 590, 646 nm.
F IGURE 8. Lithocholic acid 22 and lithocholenic acid 23.
Oxidation of the unsaturated 23 with catalyst 9 affords
products from hydroxylation in rings A, B, and D, but no
product from oxidation of the double bond in ring C, or next
to it. In the complex of 23 with 9, ring C is inside a cyclodextrin
and thus inaccessible to the porphyrin catalytic unit.
extent, but there was not such good selectivity as we saw
with substrates 6 or 11. The cyclodextrins are attached
to the porphyrin units in our catalysts by relatively
flexible links, so the geometries are not well-defined
enough for complete selectivity in the hydroxylations.
Other work is underway to improve this situation.
(30) E.g.: Winkler, J .; Coutouli-Argyropoulou, E.; Leppkes, R.;
Breslow, R. J . Am. Chem. Soc. 1983, 105, 7198.
(31) Bhacca, N. S.; Williams, D. H. Applications of NMR Spectros-
copy in Organic Chemistry: Illustrations from the Steroid Field; Holden-
Day: San Francisco, 1964.
(32) Leung, D. K.; Atkins, J . H.; Breslow, R. Tetrahedron Lett. 2001,
42, 6255-6258.
(33) (a) Bell, A. M.; Chambers, V. E. M.; J ones, E. R. H.; Meakins,
G. D.; WMuller, W. E.; Pragnell, J . J . Chem. Soc., Perkin Trans. 1
1973, 312. (b) Chambers, V. E. M.; Denny, W. A.; Evans, J . M.; J ones,
E. R. H.; Kasal, A.; Meakins, G. D.; Pragnell, J . J . Chem. Soc., Perkin
Trans. 1 1973, 1500.
Studies on androstane substrates with only one bind-
ing group, linked either at the 3 or 17 position of the
steroid, resulted in a low yielding mixture of several
compounds with no obvious major product. However, with
catalyst 9 we were able to hydroxylate lithocholic acid
22 and lithocholenic acid 23, without any added binding
group. These bind well enough34 into a single cyclodextrin
(34) Yang, Z.; Breslow, R. Tetrahedron Lett. 1997, 38, 6171-6172.
(35) Yang, Z. Ph.D. Thesis, Columbia University, 1999. This thesis
contains the X-ray crystal structure of the complex.
5064 J . Org. Chem., Vol. 67, No. 15, 2002

Catalytic Oxidations of Steroid Substrates
was obtained (>97% yield): UV/vis (H2O) 383, 405, 428, 470
(Soret band), 569, 607 nm.
28: th e Tetr a su lfon e of Ca ta lyst 7. Porphyrin 27 (20 mg,
3.8 umol) was dissolved in 5 mL of water, magnesium
monoperoxyphthalate (23.5 mg, 80% technical grade) was
added, and the reaction was stirred overnight at room tem-
perature. TLC showed complete disappearance of starting
material. Two drops of concentrated HCl was added and the
solution turned blue/green. The solution was extracted with
ether, and the aqueous layer was reduced and concentrated.
The protonated green porphyrin was lyophilized from water:
UV/vis 418 (Soret), 517, 552, 585, 639 nm in pH 7 water; IR
1405, 1282, 1153, 1073, 1030 cm-1
.
The sulfone porphyrin (∼20 mg) was dissolved in 4 mL of
anhydrous DMF and stirred under argon. Thirty-five equiva-
lents of MnCl2 (16.3 mg)) and 50 equiv of 2,6-lutidine were
added via syringe, and the solution was heated to 90 °C for
24 h and monitored by UV. The DMF was removed under
reduced pressure, and the green solid 28 was purified on a
Sephadex G75 size-exclusion column (>97% yield): UV/vis
321, 378, 399, 418, 465 (Soret), 561, 597 nm (in water).
P er flu or in a ted Ca ta lyst 9. This was first prepared by
reaction of the corresponding tetrathiol with â-cyclodextrin
6-iodide, later by reaction of the tetrakispentafluorophenylpor-
phyrin with cyclodextrin thiol. The latter procedure is analo-
gous to the one described below for compound 14, with eight
extra chlorines. For porphyrin 10: 1H NMR (500 MHz, DMF-
d7) δ -2.73 (s, 2H), 3.5-6.7 (m, 276H), 9.76 (s, 8H); 19F NMR
(282.4 MHz, DMF-d7) δ -138.6, 133.4; UV/vis (H2O) λ 412,
510, 582 nm. For manganese catalyst 9: UV/vis (H2O) λ 369,
458, 555 nm.
P er h a logen a ted Ca ta lyst 14. The perhalogenated por-
phyrin was synthesized by reaction of pentafluorobenzalde-
hyde with 3,4-dichloropyrrole in the standard way (details of
the pyrrole and porphyrin syntheses are in the Supporting
Information) and had spectroscopic data in line with those
reported.36 This porphyrin (26.9 mg, 0.0215 mmol) and K2CO3
(30.1 mg, 0.218 mmol) were dissolved in 1.6 mL of dry DMF.
After 5 min, 6-deoxy-6-mercapto-â-cyclodextrin (126 mg, 0.109
mmol) was added under argon, and the solution was stirred
at rt for 24 h. The solvent was evaporated and the residue
precipitated in acetone to give a dark green solid. The crude
product 29 was purified by gel filtration (Sephadex G-75, 20-
50 µm) with water as eluent to obtain 95 mg (77% yield) of a
brown powder after lyophilization: 1H NMR (300 MHz, DMSO-
d6/D2O ) 95:5 at 353 K) δ 5.0-4.7 (m, 28H), 4.15-4.05 (m,
4H), 3.9-3.1 (m, 244H); 19F NMR (272 MHz, DMSO-d6/D2O )
95:5 at 353 K) δ -132.6 to -133.6 (m, 8F), -139.5 to -140.5
(m, 8F); UV/vis (H2O) λmax 447 (Soret).
Then 29 (25 mg, 0.0043 mmol), MnCl2 (11.6 mg, 0.092
mmol), and 2,6-lutidine (13.8 mg, 0.129) was dissolved in 2.4
mL of dry DMF. The mixture was heated under reflux for 6 h
in air. After cooling, the solvent was evaporated and the
residue dried under vacuum overnight. The crude product was
purified by gel filtration (Sephadex G-75, 20-50 µm) with
water as eluent to obtain 23 mg (0.0039) of 14 as a green
powder: UV/vis λmax 451 (Soret), 578 nm.
Syn th eses of th e Su bstr a tes. The synthesis of 3-(4-tert-
butylphenyl)glutaric anhydride 30 will serve as a generic
synthesis of the glutaric acid linkers from the corresponding
aldehydes.
To the tetrasulfoxide porphyrin 25 (540 mg, 0.626 mmol)
in 25 mL of CH2Cl2 was added trifluoroacetic anhydride (8 mL,
57 mmol) with stirring. The mixture was gently refluxed at
40 °C for 1.5 h, the solvent was evaporated under reduced
pressure, and the product was precipitated with 25 mL of
petroleum ether. The blue/green product 26 was collected and
dried under vacuum (yield 100%): 1H NMR (300 MHz, CDCl3)
δ 5.97 (s, 8H), 8.08 (d, 8H, J ) 8.2 Hz), 8.54 (d, 8H, J ) 8.2
Hz), 8.63 ppm (s, 8H).
Four equivalents of â-cyclodextrin 6-iodide (2.27 g, 1.83
mmol) was dissolved in 100 mL of anhydrous DMF, NEt3 (4.2
mL, redistilled from CaH), and anhydrous MeOH (4.2 mL)
were added, and the solution was stirred under argon for 20
min. The cyclodextrin solution was then transferred under
argon via cannula to a flask containing porphyrin 26. The
transfer was done at a rate such that all solution was added
after 20 min. The new solution was stirred at room temper-
ature for 15 min and then heated at 60 °C for 36 h. The DMF
was then removed under vacuum, and the product was
precipitated with acetone in two portions in a centrifuge tube.
The product was then taken up several times in a minimum
amount of water and precipitated with mixtures of acetone/
MeOH/CH2Cl2. The solid was dried in a vacuum, and 2.35 g
crude product was obtained. The final product was purified
first by a reversed-phase column eluting with pure water, then
size-exclusion column (Sephadex G75) eluting with pure water,
and again with a reversed-phase column to yield 1.2 g (50%)
of pure product 27: 1H NMR (500 MHz, DMSO-d6) δ 3.2-4.2
(m, 168H), 4.3-4.65 (m, 24H), 4.75-5.1 (m, 28H), 5.6-5.9 (m,
56H), 7.83 (d, J ) 8 Hz, 8H), 8.12 (d, J ) 8 Hz, 8H), 8.92 (s,
8H); 13C NMR (125 MHz, DMF-d7) δ 35.57, 61.31, 71.65, 73.23,
73.58, 73.78, 74.10, 74.23, 82.84, 86.16, 103.12, 103.30, 103.83,
120.44, 127.44, 132.30, 135.52, 138.58, 139.25; MS (negative
ion, electrospray) 1736 (M - 3H)3-, 1302 (M - 4H)4-; UV/vis
(H2O) 422 (Soret), 523, 559, 590, 650 nm.
4-tert-Butylbenzaldehyde (24.26 g, 150 mmol) and ethyl
cyanoacetate (20.36 g, 180 mmol) in 200 mL of ethanol with 1
mL of pyrrolidine stood for 5 h at room temperature. The
ethanol was removed under reduced pressure, and the product
was extracted with 200 mL of EtOAc and 50 mL of ammonium
chloride. The EtOAc layer was washed with 50 mL of brine
and then dried over MgSO4 for several hours. The solvent was
Ca ta lyst 7. Porphyrin 27 (0.50 g, 0.0965 mmol) was
dissolved in 70 mL of anhydrous DMF and stirred under argon.
Then 35 equiv of MnCl2 (425 mg) and 50 equiv of 2,6-lutidine
(0.56 mL) was added via syringe, and the solution was heated
to 90 °C for 6 h under air and monitored by UV/vis until the
reaction was complete. The DMF was removed under reduced
pressure, and the green solid was purified on a Sephadex G75
size- exclusion column to isolate and separate the product from
MnCl2. The same procedure was repeated with another 500
mg of tetramer porphyrin and a total of 992 mg of catalyst 7
(36) Birnbaum, E.; Hodge, J .; Grinstaff, M.; Schaefer, W.; Henling,
L.; Labinger, J .; Bercaw, J .; Gray, H. Inorg. Chem. 1995, 34, 3625.
J . Org. Chem, Vol. 67, No. 15, 2002 5065

Yang et al.
and the product 32 (67% yield) was purified by column
chromatography (95/5 CH2Cl2/MeOH on SiO2): 1H NMR (500
MHz, CDCl3) δ 0.5-1,8 (m, 45H), 1.98 (m, 1H), 2.5-2.8 (m,
8H), 3.59 (m, 2H), 4.48 (m, 1H), 4.57 (br, 1H), 7.13 (d, J ) 8.0
Hz, 4H), 7.30 (d, J ) 8.0 Hz, 4H); MS (CI, NH3) 802 (M + 1 +
NH3).
The diacid derivative 32 (150 mg, 0.19 mmol), 4 equiv of
N-hydroxysuccinimide (87.5 mg, 0.76 mmol), and 4 equiv of
1-(3-diethylaminopropyl)-3-ethylcarbodiimide hydrochloride
(EDC) (146 mg, 0.76 mmol) were stirred in 30 mL of methylene
chloride overnight. The reaction mixture was quenched with
50 mL of saturated ammonium chloride solution, the organic
layer was extracted with water and dried over sodium sulfate,
and the solvent was removed under reduced pressure. The
product 33 (67% yield) was purified by column chromatography
(95/5 CH2Cl2/MeOH on SiO2): 1H NMR (500 MHz, CDCl3) δ
0.5-1.7 (m, 45H), 1.97 (m, 1H), 2.5-3.1 (m, 16H), 3.6 (m, 2H),
4.46 (m, 1H), 4.57 (br, 1H), 7.16 (d, J ) 8.2 Hz, 4H), 7.31 (d,
J ) 8.2 Hz, 4H); MS (CI, NH3) 997 (M + 1 + NH3).
The activated diester derivative 33 (180 mg, 0.19 mmol), 4
equiv of taurine (95 mg, 0.76 mmol), and 4 equiv of NEt3 were
stirred in 20 mL of anhydrous DMF overnight at room
temperature, the DMF was removed under reduced pressure,
and the product was precipitated from ∼10 mL of 2 N HCl.
The slurry was washed with acetonitrile and 1 N HCl to
produce an off-white solid. The product was taken up in water
and lyophilized to give 137 mg of substrate 6 (overall 20% yield
from androstane-3,17-diol): 1H NMR (500 MHz, DMSO-d6) δ
0.38-1.51 (m, 45H), 1.82 (m, 1H), 2.2-2.7 (m, 12H), 3.20 (m,
4H), 3.38 (m, 2H), 4.35 (m, 2H), 7.09 (d, J ) 8.1 Hz, 4H), 7.24
(d, J ) 8.1 Hz, 4H); MS (FAB) 1000 (M + 1).
Tr ip ly Lin k ed An d r osta n etr iol Su bstr a te (11) This was
prepared from the triol 8 by the sequence described above for
substrate 6: 1H NMR (500 MHz, DMSO-d6) δ 7.68 (m, 3H),
7.22 (m, 6H), 7.07 (m, 6H), 4.25 (m, 3H), 3.20 (m, 6H), 0.62 (s,
3H), 0.46 (s, 3H), 3.4-2.28 (21H), 1.85-0.30 (23H).
removed under reduced pressure, and the product was taken
on to the next step without further purification.
To the above product in 155 mL of EtOH were added
Meldrum’s acid (23.8 g) and NaOH (6.9 g in 75 mL water).
The mixture was stirred at room temperture for 3 h and at 70
°C overnight. Ethanol was removed under reduced pressure,
and the product was dissolved in 170 mL of concd HCl and
heated under reflux for 10 h. The suspension was filtered, and
the HCl solution was extracted with EtOAc. The organic
extracts were combined with the filtered solid, the solvent was
removed under reduced pressure, the product was then
suspended in 150 mL of ethylene glycol with NaOH (35.6 g),
and the mixture was heated to 150 °C for 24 h.
After the solution was cooled, 100 mL of ice-water was
added, and the solution was then acidified to pH 1-2 with
concd HCl and cooled in an ice bath. The solid crude product
was filtered and dissolved in 100 mL of 1 N NaOH. The basic
solution was extracted with ether to remove most of the
impurities, and the aqueous layer was again acidified to pH
1-2 with concd HCl to precipitate the product 3-(4-tert-
butylphenyl)glutaric acid 31 in an ice bath. This purification
scheme was done up to three times to ensure complete removal
of impurities. The final product was filtered from the aqueous
acid and the washed with 95-98% petroleum ether/ether
solution to produce 17 g of 31 as an off-white solid: 1H NMR
(200 MHz, DMSO-d6) δ 1.2 (s, 9H), 2.5 (m, 4H), 3.38 (q, 1H),
7.15 (d, J ) 8.4 Hz, 2H), 7.28 ppm (d, J ) 8.4 Hz, 2H); 13C
NMR (75 MHz, DMSO-d6) δ 31.3, 34.2, 37.5, 40.3, 125.1, 127.2,
140.4, 148.6, 173.0.
The syntheses of the unsymmetrically doubly linked an-
drostane derivatives 17, 18, and 19 are described in Support-
ing Information.
Typ ica l Con d ition s for Hyd r oxyla tion s of th e Ster oid
Su bstr a tes a n d th e HP LC An a lysis Meth od s. A typical
experiment: Substrate 6 (20 mg), x mol % catalyst 7 or 9, and
10 equiv (with respect to substrate) of pyridine were dissolved
in 20 mL of deionized water and stirred in a closed vessel for
10 min at room temperature in the dark. A solution of 5 equiv
of PhIO in MeOH (30 mg/mL) was added to the solution via
syringe in 0.2-0.3 mL increments over a period of 20 min.
Another 5 equiv of PhIO was added in the same manner over
another 20 min. After the solution was stirred at room
temperature for ∼2 h, excess Na2S2O3 solid was added followed
by 10 mL of 3 N NaOH to hydrolyze the esters. The solution
was stirred overnight, after which time the solution was
extracted with 3 × 20 mL of EtOAc. The combined organic
extracts were then washed with brine and dried over Na2SO4.
The solvent was removed under reduced pressure, and the
crude product mixture (∼0.02 mmol of total steroid) was taken
up in 10 mL of distilled pyridine.
Freshly distilled benzoyl chloride (0.5 mL, ∼430 equiv with
respect to steroid) was added to the crude pyridine solution
along with 1 mg of 4-dimethylaminopyridine (DMAP). The
solution was then purged with argon, shaken for 10 min, and
put in an 80 °C bath for 20-25 min. The reaction mixture was
poured into a separatory funnel containing 50 mL of 0.1 N
HCl and 50 mL of 90:10 hexanes/ethyl acetate and then
washed once with 20 mL of 0.1 N HCl and once with 20 mL of
saturated NaHCO3. The organic layer was dried over Na2SO4
and brought to dryness under reduced pressure to give a brown
oil. A small aliquot of this oil was taken up in 1 mL of
2-propanol and analyzed by HPLC.
Diacid 31 (5 g) was dissolved in 100 mL of acetic anhydride,
the solution was heated at reflux for several hours, and after
all starting material disappeared (by TLC) the solvent was
removed under reduced pressure. The product 3-(4-tert-but-
ylphenyl)glutaric anhydride 30 was washed with
a 10:1
petroleum ether/ether solution until the brown color disap-
peared. An off-white solid was collected (4.6 g) and vacuum-
dried: 1H NMR (200 MHz, DMSO-d6) δ 1.27 (s, 9H), 3.0 (m,
4H), 3.5 (m, 1H), 7.2 (d, J ) 8.3 Hz, 2H), 7.39 ppm (d, J ) 8.3
Hz, 2H).
5R-Androstane-3,17-diol (200 mg), 3-(4-tert-butyl)glutaric
anhydride 30 (337 mg), and 20 mg of p-toluenesulfonic acid
were heated at reflux in 25 mL of toluene under argon
overnight. The solvent was removed under reduced pressure,
Authentic samples of the androstanediols and -triols were
quantitatively converted to their benzoate esters using the
same method described for derivatization of the crude mix-
5066 J . Org. Chem., Vol. 67, No. 15, 2002

Catalytic Oxidations of Steroid Substrates
tures. Further details of the analytical method are described
in the Supporting Information.
Typical conditions for hydroxylations of lithocholic 20 and
lithocholenic 21 acids and GC analysis methods are described
in the Supporting Information.
were consistent with reported values.31 NOESY indicated that
the C5 and C7R protons moved downfield significantly.
3-Keto-5â-ch ola n ic Acid . All cholanic and cholenic acid
products had the correct masses: 1H NMR (CDCl3, 500 MHz)
δ 1.02 (s, 3H), 0.923 (m, 3H), 0.683 (s, 3H). The product was
compared and confirmed against authentic product obtained
from Steraloids, Inc.
Sp ectr oscop ic Ch a r a cter istics of Ster oid P r od u cts. 5r-
An d r osta n e-3â,17â-d iol was purchased: 1H NMR (CDCl3,
500 MHz) δ 3.62 (t, 1H), 3.55 (m, 1H), 0.818 (s, 3H), 0.732 (s,
3H).
3-Keto-5â-ch ol-11-en ic a cid : 1H NMR (CDCl3, 400 MHz)
δ 1.02 (d, J ) 6 Hz, 3H), 0.98 (s, 3H), 0.768 (s, 3H).
5â-Ch ol-11-en ic a cid -3r,1â-d iol: 1H NMR (CDCl3, 500
MHz) δ 4.05 (br s, 1H), 4.02 (m, 1H), 0.856 (s, 3H), 0.737 (s,
3H). The C1-H and C3-H have a common COSY coupling to
the C2-H’s.
1
5r-An d r osta n e-3â,6r,17â-tr iol (8): H NMR (CDCl3, 500
MHz) δ 3.66 (t, 1H), 3.61 (m, 1H), 3.44 (dt, 1H), 0.834 (s, 3H),
0.738 (s, 3H); CI-MS m/z ) 326 (M + 1 + NH3). The NOE
spectrum indicates a strong signal between the C6-H to the
C19-Me. Triol 8 was compared with an authentic sample
obtained from Steraloids, Inc.
5â-Ch ol-11-en ic a cid -3r,6r-d iol: 1H NMR (CDCl3, 500
MHz) δ 4.11 (m, 1H), 3.65 (m, 1H), 0.837 (s, 3H), 0.728 (s, 3H).
NOE observed between the C6-H and the C19-Me.
5â-Ch ol-11-en ic a cid -3r,6â-d iol: 1H NMR (CDCl3, 500
MHz) δ 3.83 (br m, 1H), 3.65 (m, 1H), 1.08 (s, 3H), 0.773 (s,
3H).
5r-An d r osta n e-3â,7â,17â-tr iol (18): 1H NMR (CDCl3, 500
MHz) δ 3.62 (t, 1H), 3.58 (m, 1H), 3.38 (m, 1H), 0.849 (s, 3H),
0.762 (s, 3H); CI-MS m/z ) 326 (M +1 + NH3). The new CH-
OH and methyl shifts compared to the starting diol are
consistent with hydroxylation at C7.31 The COSY spectrum
indicated 3 neighboring protons for the new CH peak. NOESY
indicated two cross-peaks with the C7R-H, presumably to the
C5-H and C14-H.
5â-Ch ol-11-en ic a cid -3r,15r-d iol: 1H NMR (CDCl3, 500
MHz) δ 4.04 (m, 1H), 3.65 (m, 1H), 0.887 (s, 3H), 0.783 (s, 3H).
NOE observed between the C15-H and the C18-Me.
5â-Ch ol-11-en ic a cid -3r,15â-d iol: 1H NMR (CDCl3, 500
MHz) δ 4.26 (m, 1H), 3.65 (m, 1H), 1.05 (s, 3H), 0.922 (s, 3H).
5r-An d r osta n e-3â,15r,17â-tr iol (19): 1H NMR (CDCl3,
500 MHz) δ 4.06 (m, 1H), 3.87 (t, 1H), 3.61 (m, 1H), 0.842 (s,
3H), 0.755 (s, 3H); CI-MS m/z ) 326 (M + 1 + NH3). The new
CH-OH and methyl shifts compared to the starting diol are
consistent with hydroxylation at C15.31 The COSY spectrum
indicated three neighboring protons for the new CH, two of
which were the 16 protons as determined by a common COSY
to the C17-H. NOESY revealed a strong cross-peak of the
C15-H to the C18-Me, indicating that the hydroxyl group is
15R.
Ack n ow led gm en t. This work was supported by the
NIH, NSF, and EPA. J .Y. acknowledges support from
a Bristol-Myers Squibb Graduate Fellowship and an
EPA NCERQA STAR Graduate Fellowship. B.G. ac-
knowledges support from a NATO postdoctoral fellow-
ship. The authors thank Anneka Lenssen for her help.
Su p p or tin g In for m a tion Ava ila ble: Additional experi-
mental details. This material is available free of charge via
the Internet at http://pubs.acs.org.
5r-An dr ostan e-3â,6r,9r,17â-tetr aol (12): 1H NMR (CDCl3,
500 MHz) δ 3.73 (t, 1H), 3.58 (m, 1H), 3.43 (dt, 1H), 0.959 (s,
3H), 0.753 (s, 3H); CI-MS m/z ) 342 (M + 1 + NH3), 323
(M - 1, neg). The chemical shifts of the angular methyl groups
J O020174U
J . Org. Chem, Vol. 67, No. 15, 2002 5067