Tethered aminohydroxylation: Dramatic improvements to the process

Article abstract of DOI:10.1021/ol070430v

Changing the identity of the N leaving group on a hydroxylamine-based reoxidant gives a dramatic improvement to the tethered aminohydroxylation reaction. Using OCOC₆F₅ as a leaving group means that only 1 mol % of osmium is required and yields as high as 98% can be obtained. Acyclic homoallylic alcohols were substrates considered too unreactive for effective use in the tethered aminohydroxylation reaction; improved reaction conditions mean that they have now become viable substrates for oxidation.

Full text of DOI:10.1021/ol070430v

ORGANIC
LETTERS
2007
Vol. 9, No. 9
1725-1728
Tethered Aminohydroxylation: Dramatic
Improvements to the Process
Timothy J. Donohoe,*,† Carole J. R. Bataille,† William Gattrell,‡
Johannes Kloesges,† and Emilie Rossignol†
Department of Chemistry, UniVersity of Oxford, Chemistry Research Laboratory,
Mansfield Road, Oxford OX1 3TA, U.K., and Prosidion Ltd., Watlington Road,
Oxford OX4 6LT, U.K.
timothy.donohoe@chem.ox.ac.uk
Received February 20, 2007
ABSTRACT
Changing the identity of the
N
leaving group on
a
hydroxylamine-based reoxidant gives
a
dramatic improvement to the tethered
aminohydroxylation reaction. Using OCOC6F5 as a leaving group means that only 1 mol % of osmium is required and yields as high as 98%
can be obtained. Acyclic homoallylic alcohols were substrates considered too unreactive for effective use in the tethered aminohydroxylation
reaction; improved reaction conditions mean that they have now become viable substrates for oxidation.
The widespread occurrence of amino alcohols and amino
acids in biologically active molecules and natural products
has led to the need for a simple, efficient, and reliable
methodology for the introduction of such functional groups.
The pioneering asymmetric aminohydroxylation (AA) reac-
tion, as developed by Sharpless, is a powerful method for
the stereospecific preparation of vicinal amino alcohols using
substoichiometric potassium osmate in the presence of a
nitrogen source (typically a carbamate) and tert-butyl hy-
pochlorite as a reoxidant.1 We have extended the scope of
this method by tethering the carbamate unit to an allylic
alcohol, thus constituting a tethered aminohydroxylation
(TA). This approach solves one of the drawbacks of the AA
reaction, which is control of regioselectivity during the
oxidation of unsymmetrical alkenes.2
moderate yields and with total regioselectivity (Scheme 1).
These results were encouraging, and further investigation was
undertaken to improve the yields. We observed that chlorina-
tion of the alkene unit was a competing side reaction in the
TA reaction under protocol A and is partly responsible for
lowering the yield. Therefore, a screen of alternative reoxi-
dants for the TA reaction was conducted, focusing on
Scheme 1
Initially, allylic carbamates were oxidized with t-BuOCl
using Sharpless’ original conditions3 (named TA protocol
A hereafter), and the desired products were obtained in
† University of Oxford.
‡ Prosidion.
(1) For reviews, see: Kolb, H. C.; Sharpless, K. B. In Transition Metals
for Organic Synthesis; Beller, M., Bolm, C., Eds.; Wiley: New York, 1998;
Vol. 2, pp 243. Bodkin, J. A.; McLeod, M. D. J. Chem. Soc., Perkin Trans.
1 2002, 2733.
(2) Donohoe, T. J.; Johnson, P. D.; Cowley, A.; Keenan, M. J. Am. Chem.
Soc. 2002, 124, 12934.
10.1021/ol070430v CCC: $37.00
© 2007 American Chemical Society
Published on Web 03/28/2007

replacement of the chlorine of the N-halocarbamate salt that
is generated in situ during oxidation with t-BuOCl and
NaOH.
Scheme 2
This study led to the discovery of N-mesitylsulfonyloxy
derivatives as valuable reoxidants for osmium. In these cases,
the N-Cl unit had been replaced by a N-O-SO2Mes group,
which was introduced before the aminohydroxylation rather
than being formed in situ. Allylic and homoallylic substrates
were successfully oxidized using the novel and chlorine-
free TA (named hereafter TA protocol B); the corresponding
products were usually obtained in good yields.4 Although
this new protocol was significantly better than the original
TA reaction based on t-BuOCl, the reaction sometimes
proved capricious; that is, lower yields were obtained for
no apparent reason on some substrates (Scheme 1). In
addition, the formation of the TA precursors from the
corresponding alcohols could be problematic, leading to
moderate yields and fairly unstable compounds. It was,
therefore, decided to investigate the reaction further and to
especially focus our attention on these particular issues, as
they constitute an obstacle to the widespread utilization of
the TA reaction in the chemistry community.
pentafluorobenzoyl substrate 7d was subjected to the new
TA protocol, the desired oxazolidinone 8 was obtained in
excellent and reproducible yield. It is noteworthy that the
amount of potassium osmate could be significantly reduced
to 1 mol % without lowering the yields; this change was
not always possible before, especially with protocol A. These
improvements render the novel TA protocol C very attractive
as a synthetic route to vicinal amino alcohols.
Although encouraging results were obtained with the
O-SO2Mes reoxidant, no trend emerged as to which sub-
strates would work well, or poorly, in the reaction. It was,
therefore, decided to screen a wider range of leaving groups
on the carbamate nitrogen. Cinnamyl alcohol 5 was chosen
as a test substrate for this investigation as this alcohol is
commercially available and a good example of a class of
compounds (primary alcohol, trans-alkene) which were less
than ideal under previous protocols.5
Table 1. Effect of Acyl-Based Leaving Groups on the Yield of
the TA
yield of 8
entry
R
yield of 7a-f
(protocol C)
7a
7b
7c
7d
7e
7f
Me
81%
85%
81%
95%
90%
94%
52%
73%
86%
86%
86%
0%
Previously, the preparation of hydroxycarbamates, from
the corresponding alcohol, was performed using CDI and
hydroxylamine hydrochloride in acetonitrile in the presence
of imidazole; this method sometimes led to moderate yields
due to the formation of byproducts. A judicious switch to
pyridine solvent allowed the reduction of byproducts and
led to dramatic increases in the yields. For example, cinnamyl
alcohol 5 was reacted with CDI in pyridine, followed by
the addition of hydroxylamine hydrochloride, to afford
hydroxycarbamate 6 in excellent yields (Scheme 2). The
resulting hydroxycarbamate 6 was then treated with several
acid chlorides, in ether, to yield the corresponding O-
derivatized hydroxycarbamates 7, which were subjected to
the new TA conditions (named TA protocol C hereafter), to
test their potential as reoxidants (Table 1).
2,4,6-Cl3C6H2
2,4,6-Me3C6H2
C6F5
p-ClC6H4
t-Bu
To confirm the potential of this improved procedure, it
was decided to oxidize a series of allylic alcohol derivatives
using the new protocol C (Scheme 3). The alcohols 11-14
were treated with CDI in pyridine followed by the addition
of hydroxylamine hydrochloride to afford the hydroxycar-
bamate intermediates, which were then converted to the
(7) During the screening of nitrogen leaving groups, we discovered that
often the TA adduct 8 was obtained with the carbamate 10 as a byproduct.
It is interesting that the ratio of TA product 8 to carbamate 10 depends on
the nature of the R group positioned on the O-acyl group of the
hydroxycarbamate derivatives. The formation of the carbamate was surpris-
ing. It is proposed that the isolation of 10 could be due to two factors.
First, the influence of a particular R group on the solubility of the substrate
may play a role in determining the outcome of the reaction. Second, the
pKa of the conjugate acid of the leaving group (RCOO-) may be important
in determining the fate of the starting material. For example, the O-PFB
moiety (R ) C6F5) is one of the best leaving groups examined here (and
this gave no carbamate in the TA reaction) whereas the O-Piv group (see
9, R ) t-Bu) is the worst leaving group studied. This led to the formation
of carbamate 10 in quantitative yield.
Interestingly, the O-pentafluorobenzoyl group (O-PFB)
emerged as a superior reoxidant for the TA reaction, giving
a dramatically improved yield (see entry 7d).6,7 When
(3) For example, see: Li, G.; Angert, H. H.; Sharpless, K. B. Angew.
Chem., Int. Ed. Engl. 1996, 35, 2813. Laxma Reddy, K.; Dress, K. R.;
Sharpless, K. B. Tetrahedron Lett. 1998, 39, 3667. Laxma Reddy, K.;
Sharpless, K. B. J. Am. Chem. Soc. 1998, 120, 1207.
(4) Donohoe, T. J.; Chughtai, M. J.; Klauber, D. J.; Griffin, D.; Campbell,
A. D. J. Am. Chem. Soc. 2006, 128, 2514. For the use of similar reagents
for aziridination and C-H insertion, see: Lebel, H.; Huard, K.; Lectard,
S. J. Am. Chem. Soc. 2005, 127, 14198.
(5) Donohoe, T. J.; Helliwell, M.; Johnson, P. D.; Keenan, M. Chem.
Commun. 2001, 2078.
(6) Kloesges, J. Part II thesis; Oriel College: Oxford, 2006.
1726
Org. Lett., Vol. 9, No. 9, 2007

Scheme 3
Scheme 4
corresponding pentafluorobenzoyl derivatives 15-18 in good
to excellent yields. The PFB substrates 15-18 were oxidized
using protocol C to afford the desired products 2 and 19-
21 in good to very good yields and with good diastereo-
selectivity where applicable.
outcome of the TA reaction of 25; this gave oxazinanone 4
in only 23% yield using protocol A and in 40% yield using
protocol B. However, using protocol C (1 mol %), this yield
was improved dramatically. The high level of 1,3-acyclic
stereocontrol in favor of the anti diastereoisomer is also
noteworthy.
The identities of oxazolidinones 2 and 19-21 were
confirmed by comparison with the analytical data of authentic
samples previously synthesized using protocol A and/or B.2,5,8
Pleasingly, all the yields obtained with protocol C were
significantly higher than the ones afforded by the two
previous methods, using only 1 mol % of potassium osmate
instead of the 4 mol % utilized exclusively in protocol A
and for the most part with protocol B.
Relative stereochemistries in this series were confirmed
by X-ray crystallography for oxazinanone 4 and by NOE
experiments for oxazinanones 28 and 29. None of these
acyclic homoallylic alcohols gave synthetically useful yields
in the TA reaction using the previous two procedures.
We also propose here a simple model that could serve as
an explanation for the stereoselectivity observed in this
system. On the basis of the lack of asymmetric induction in
the TA reaction in the presence of a chiral ligand, we have
previously suggested that the TA reaction works through the
second catalytic cycle.5,9 So, we assume that the diastereo-
selectivity obtained above arises from an intermediate such
as A or B, where the osmium atom has already undergone
addition across an alkene unit and then been reoxidized
(Figure 1). To rationalize the stereochemical outcome of the
We then attempted to apply this novel methodology to
homoallylic alcohol derivatives that have previously been
unsuitable for a TA reaction under both earlier sets of
conditions. Pentafluorobenzoyl derivatives 25-27 were
prepared in good to excellent yields from the corresponding
homoallylic alcohols (()-22-24 via the previously described
method (Scheme 4) and then submitted to TA protocol C,
affording the desired oxazinanones 4, 28, and 29 in good to
excellent yields and good to excellent diastereoselectivity.
It is noteworthy that, again, all the reactions were performed
with only 1 mol % of potassium osmate. Consider the
(9) Wai, J. S. M.; Marko´, I.; Svendsen, J. M.; Finn, M. G.; Jacobsen, E.
N.; Sharpless, K. B. J. Am. Chem. Soc. 1989, 111, 1123. Wu, P.; Hilgraf,
R.; Fokin, V. V. AdV. Synth. Catal. 2006, 348, 1079. Mun˜iz, K.; Almodovar,
I.; Streuff, J.; Nieger, M. AdV. Synth. Catal. 2006, 348, 1831.
(8) Donohoe, T. J.; Johnson, P. D.; Keenan, M.; Pye, R. J. Org. Lett.
2004, 6, 2583.
Org. Lett., Vol. 9, No. 9, 2007
1727

are difficult to understand because the less selective deriva-
tive 26a could readily put the OEt group in an equatorial
position on the six-membered ring. It may be that this type
of conformation is relatively unreactive because the equato-
rial ethoxy group is placed close to antiperiplanar to the
alkene where it can deactivate it by stereoelectronic effects.11
To conclude, we have shown that the nature of the leaving
group attached to the N-O reoxidant for aminohydroxylation
is crucial in promoting an efficient reaction. Positioning the
pentafluorobenzoyl group on the oxygen turns an acyl
hydroxylamine into an extremely efficient reoxidant capable
of giving yields as high as 98% for the tethered amino-
hydroxylation reaction, all with just 1 mol % of osmium
catalyst. Each of the major classes of substrates has been
shown to be compatible with this change, and acyclic
homoallylic alcohols have, for the first time, become viable
compounds for the TA reaction, proceeding with high levels
of 1,3-asymmetric induction.
Figure 1. Transition state model for 1,3-anti diastereoselectivity.
oxidation, we make the assumption that the OsdN-C
linkage is a linear one without much change in shape in the
transition structure.8,10 Examination of molecular models then
shows that a boatlike conformation A (with an equatorial R
group) allows the best overlap between the oxidant and the
π-bonds of the alkene (this gives rise to 1,3-anti diastereo-
selectivity). The corresponding chairlike transition state B
(which would give the syn isomer) suffers because this
arrangement puts a larger distance between the orbitals of
the alkene and the imido-osmium complex and is, therefore,
disfavored.
Acknowledgment. We would like to thank Prosidion (a
subsidiary of OSI Pharmaceuticals, Inc.) and the Leverhulme
Trust for funding this project. AstraZeneca, Merck, Novartis,
Pfizer, and the ESPRC are also thanked for support.
Supporting Information Available: Copies of 1H NMR
spectra and detailed spectroscopic data for all new com-
pounds and representative experimental procedures. This
material is available free of charge via the Internet at
http://pubs.acs.org.
The matching and mismatching effects of a second
stereogenic center on the carbon backbone (see 28a and b)
OL070430V
(10) Thomas, S.; Lim, P. J.; Gable, R. W.; Young, C. G. Inorg. Chem.
1998, 37, 590. Of course, the C-N-Os linkage is not linear in the product
of the reaction.
(11) Haller, J.; Strassner, T.; Houk, K. N. J. Am. Chem. Soc. 1997, 119,
8031. For a view of electronic effects on the reaction of imido-osmium
species, see: Mun˜iz, K. Eur. J. Org. Chem. 2004, 2243.
1728
Org. Lett., Vol. 9, No. 9, 2007