Ligand-dependent reversal of facial selectivity in the asymmetric dihydroxylation

Full text of DOI:10.1021/jo961189t

7
978
J . Org. Chem. 1996, 61, 7978-7979
Liga n d -Dep en d en t Rever sa l of F a cia l
Selectivity in th e Asym m etr ic
Dih yd r oxyla tion
1
Koen P. M. Vanhessche and K. Barry Sharpless*
Department of Chemistry, The Scripps Research Institute,
1
0550 North Torrey Pines Road, La J olla, California 92037
Received J une 24, 1996
Though the mechanism of even the “simple” reaction
2
of osmium tetraoxide with an olefin remains uncertain,
there has been considerable controversy over the mech-
anism of the process when a chiral cinchona alkaloid
ligand is also involved. While it is certainly risky to
F igu r e 1. Modified mnemonic for PHAL and PYR-ligands
with mono- and 1,1-disubstituted olefins.
3
build the “superstructure” of a mechanism on such an
insecure foundation, the extensive use of the asymmetric
dihydroxylation (AD) process4 has fueled substantial
interest and speculation on this topic. On the basis of a
large body of enantioselectivity data, an empirical, but
reliable, predictive mnemonic has evolved.3
ity is maintained. With the PYR ligand the ee drops
dramatically from R-methylstyrene (69% ee) to R-ethyl-
styrene (20% ee), and beginning with R-propylstyrene
(-16% ee), reversed selectivities are observed, reaching
-35% ee for R-hexylstyrene. In retrospect, these results
are anticipated by the known poor selectivity (80% ee)
for styrene and the good selectivity (89% ee) for 1-hexene
2
a
This device has proven suitable for interpreting most
of the AD results using both the phthalazine spacer
with the (DHQD) -PYR ligand (cf. 97% ee and 80% ee,
5
(
PHAL and related “1,4-substituted spacers” ) and the
respectively, with the (DHQD) -PHAL ligand).
2
9
pyrimidine (PYR) spacer. For the present study, the
most important feature of the AD mechanism to appreci-
ate is the “binding pocket” phenomenon. We have
established the existence of a binding pocket effect by
kinetic studies,3 and despite the controversy over its
location, both camps3 support its existence. In our
mnemonic this binding pocket (Figure 1) resides in the
SW quadrant. With PHAL ligands, this pocket is a
Entries 7-11 of Table 1 show the effect of R-branching
in both acyclic and cyclic systems. Changing from
isopropyl to tert-butyl (Table 1, entries 7 and 8) causes a
remarkable drop in ee with the PHAL ligand, from 82 to
8% ee. Calculations reveal that with at least one
hydrogen at the point of attachment of the alkyl group
a
[e.g., (Me) HC-, Table 1, entry 7] the phenyl can still be
2
positioned nicely in the PHAL binding pocket. However,
“
magnet” for flat aromatic groups, whereas with PYR
when this last hydrogen becomes a methyl (e.g., (Me) C-,
3
ligands, aliphatic groups are preferentially attracted to
the “SW binding pocket”. Taken together with the “rule”
that the group cis to the substituent in the binding pocket
is ideally a hydrogen (the SE quadrant is sterically the
most crowded),3 these preferences lead to an interesting
prediction for AD applications involving R-alkylstyrenes
Table 1, entry 8), rotation of the phenyl group, in relation
to the rest of the structure, is restricted to conformers
that allow only very poor presentation of the phenyl to
the binding pocket.3
a
a
High enantioselectivities were obtained for the exo-
methylene substrates (Table 1, entries 9 and 11, 95% and
92% ee, respectively) with PHAL, and no reversal oc-
curred with PYR. Introducing gem-dimethyl groups in
the allylic position (Table 1, entry 10) produces a small
drop for PHAL (to 82% ee) and a strong reversal of facial
selectivity for PYR (to -59% ee). In other words, these
fairly rigid cyclic cases mainly follow the same trends as
the acyclic analogs. The informative exception here is
that, unlike entry 8, entry 10, being cyclic, still fits fairly
well into the PHAL binding pocket. We attribute the
origin of this difference to the fact that in the cyclic cases
the key torsional angles are locked close to the favorable
(Figure 1): Namely, that for a given pseudoenantiomer
of the alkaloid [e.g., dihydroquinidine (DHQD)] the PHAL
and PYR heterocyclic spacers should give opposite enan-
tiofacial selectivity. The systematic study presented in
Table 1 reveals that this prediction is borne out by
experiment.6 The phenomenon was discovered inde-
pendently by Krysan, and his recent report prompted
us to publish our study.
All results for the series of R-aliphatic substituted
styrenes in Table 1 were obtained using DHQD-based
ligands. The enantioselectivities observed with the
PHAL ligand drop gradually with increasing chain length
7
1
0
conformation needed to fit into the PHAL binding pocket.
(
Table 1, entries 1-6), but the expected π-facial selectiv-
The patterns that emerge for the R-cycloalkyl-substi-
tuted series (Table 1, entries 12-15) are the same as
those for their acyclic alkyl analogs (Table 1, entries 3-6).
The PHAL ligand exhibits a slight decline in enantio-
(1) Current address: Firmenich, S. A., Research Laboratory, 1, route
des J eunes, CH-1211 Geneva, Switzerland.
(2) A [3 + 2] mechanism was originally proposed by B o¨ seken for
reactions of permanganate and adopted by Criegee for osmium
tetraoxide: (a) B o¨ seken, J . Recl. Trav. Chim. 1922, 41, 199. For the
(6) A general procedure for the AD of R-alkylstyrenes is given in
the Supporting Information.
[2 + 2] mechanism, see: (b) Sharpless, K. B.; Teranishi, A. Y.; B a¨ ckvall,
J .-E. J . Am. Chem. Soc. 1977, 99, 3120. (c) Tomioka, K.; Nakajima,
M.; Koga, K. J . Am. Chem. Soc. 1987, 109, 6213. (d) Gable, K. P.;
J uliette, J . J . J . J . Am. Chem. Soc. 1995, 117, 955. (e) Gable, K. P.;
Phan, T. N. J . Am. Chem. Soc. 1994, 116, 833. (f) J ørgensen, K. A.;
Schiøtt, B. Chem. Rev. 1990, 90, 1483.
(7) Krysan, D. J . Tetrahedron Lett. 1996, 37, 1375.
(8) Related reversals of enantiofacial selectivity have also been
observed for R-arylstyrenes; Loren, S.; Sharpless, K. B. unpublished
results.
(9) Crispino, G. A.; J eong, K.-S.; Kolb, H. C.; Wang, Z.-M.; Xu., D.;
Sharpless, K. B. J . Org. Chem. 1993, 58, 3785.
(3) (a) Kolb, H. C.; Andersson, P. G.; Sharpless, K. B. J . Am. Chem.
Soc. 1994, 116, 1278. (b) Norrby, P.-O.; Becker, H.; Sharpless, K. B. J .
Am. Chem. Soc. 1996, 118, 35. (c) Corey, E. J .; Noe, M. C. J . Am. Chem.
Soc. 1996, 118, 319 and references cited therein.
(10) We have also been able to rationalize the trends in facial
selectivity for R-alkylstyrenes using molecular mechanics calculations
(see ref 3b for parameters). This approach is of course much more
complicated than the simple “binding pocket” mnemonic, but it is
especially informative for the exceptional cases 8 and 16 in Table 1:
Becker, H.; Loren, S.; Vanhessche, K.; Sharpless, K. B. Unpublished
results.
(
4) Kolb, H. C.; VanNieuwenhze, M. S.; Sharpless, K. B. Chem. Rev.
994, 94, 2483.
5) Becker, H.; King, S. B.; Taniguchi, M.; Vanhessche, K. P. M.;
Sharpless, K. B. J . Org. Chem. 1995, 60, 3940.
1
(
S0022-3263(96)01189-9 CCC: $12.00 © 1996 American Chemical Society

Communications
J . Org. Chem., Vol. 61, No. 23, 1996 7979
Ta ble 1a En a n tioselectivities,b Con figu r a tion s,c,d a n d Sign s of Op tica l Rota tion s of th e Diols fr om th e AD of
r-Alk ylstyr en es w ith (DHQD)2-P HAL a n d (DHQD)2-P YR a s th e Liga n d s
a
The isolated yields of diols were 85-95%, except for entries 8, 10, and 16 (20-40%). b Enantiomeric excesses were determined by
c
HPLC analysis of the diols or their respective derivatives; see the supporting information. The absolute configurations of the diols were
determined by comparison of their optical rotations with literature values; see the supporting information. d The assignment of “negative”
ee values is just a convenient device for making the reversal of face selectivity obvious at a glance. As a further aid, the “negative” values
are also bolded.
selectivity down the series, whereas the PYR ligand gives
reversed enantiofacial selectivities and with consistently
higher ee values than seen for their acyclic counterparts
negative “in-close” effects. These latter steric effects
must be very similar for the two substituents, yet are
calculated to be slightly larger for tert-butyl than for
-1
(e.g., Table 1, entries 3-6). These results clearly support
1-adamantyl (i.e., A value for tert-butyl ca. 0.2 kcal mol
the preference of the PHAL pocket for aromatic groups
and of the PYR pocket for aliphatic groups, especially
those with R-branching. Earlier evidence3 suggests that
both pockets benefit from a hydrophobic effect, but the
latter must be the dominant factor for the PYR pocket.
Τhe result in Table 1 that stands out from all the
others is the dramatic switch in enantiofacial selectivity
for R-(1-adamantyl)styrene using the PHAL ligand (first
column, entry 16, cf. entries 12-15). At first, the only
thing certain was that the PHAL and PYR ligands had
given the same enantiomer. We believed that it was the
PYR case that had switched faces and given the (R)-diol,
perhaps, because in this sterically demanding case, the
phenyl group had for some reason replaced the alkyl
group in the PYR-binding pocket. However, the absolute
configuration of the diol was then proven to be S by an
X-ray crystal structure determination on the mono-(1S)-
camphanic ester derivative.14 This result established
that it was the adamantyl substituent that occupied the
SW quadrant in both cases (i.e., with PYR and PHAL
ligands). In hindsight, the “strange” results in entries 8
and 16 (Table 1) are consistent with the good results for
larger than for 1-adamantyl).12 Continuing the analysis
along these lines, one notes that the increase from 64%
ee to 87% ee between the tert-butyl and the adamantyl
cases (both performed at 0 °C) represents an increase in
a
q
-1
∆∆G of ca. 0.6 kcal mol , which leads to the, admittedly,
highly speculative conclusion that the observed ee in-
crease could be due to small steric (1/3) and hydrophobic
(2/3) effects acting in concert.
Ack n ow led gm en t. K.P.M.V. thanks the Belgian
National Fund for Scientific Research and NATO for a
scholarship. We are grateful to the National Institutes
of Health (GM 28384) and to the W. M. Keck Foundation
for financial support. We also wish to thank Pui Tong
Ho, Stefan Loren, and Heinrich Becker for helpful
discussions and Stefan Immel for calculation of the A
values for the tert-butyl and 1-adamantyl substituents.
Su p p or tin g In for m a tion Ava ila ble: A general experi-
mental procedure for the AD reactions, characterization data
for all the diols, as well as the methods for their ee determi-
nation, optical rotation, and absolute configuration, and an
ORTEP for the mono-(1S)-camphanate ester of the diol derived
from R-(1-adamantyl)styrene (entry 16, Table 1) (10 pages).
1
-adamantylethylene using either (DHQD)
2
-PHAL (87%
ee, (R)-diol) or (DHQD)
2
-PYR (97% ee, (R)-diol)11 and,
J O961189T
above all, with the massive drop in ee between entries 7
and 8 (Table 1) with the PHAL ligand (rationale for this
drop, vide supra).
(12) The tert-butyl and adamantyl substituents are generally re-
garded as having similar steric demands in the regions of space where
their structures are homologous (i.e., referred to here as “in-close”).
Yet, the calculated (MacroModel 5.0) A valuessi.e., the free energy
differences between the equatorial and axial conformations for sub-
stituted cyclohexanes (tert-butyl: calculated 5.0 kcal mol- , experi-
The recent11 87% ee result for 1-adamantylethylene
using (DHQD) -PHAL requires further explanation. It
2
seemed out-of-line since tert-butylethylene was known to
give only 64% ee under identical conditions,9 and a
similar branched alkyl substituent that, like 1-adaman-
tyl, is much larger “at a distance,” might have been
expected to result in an even poorer ee. At present we
can only speculate on causes for this “adamantyl phe-
nomenon” (i.e., counter to trend for the PHAL ligand,
wherein ee is poor when a tert-alkyl group is the best
1
-1 13
-1
mental 4.9 kcal mol
;
1-adamantyl: calculated 4.8 kcal mol )s
indicate slightly lower “in-close” steric demands for the 1-adamantyl
substituent. If these small calculated differences are real, it is
interesting to speculate on the origin of the effect. The adamantane
structure is rigid and compact, and significant bond angle variations
are impossible. If one now imagines the creation of the tert-butyl group
by formal hydrogenolysis of the three appropriate C-C bonds in
adamantane, what remains, in place of the very polycyclic framework
that made deformations a nonissue, are three new C-H bonds, one
on each of the methyls of what is now a tert-butyl group. Hence, cyclic
constraint has in a sense been exchanged for nonbonded repulsion in
the acyclic analog, and even a very slight spreading of the methyls
“offering” available for the binding pocket). Perhaps it
is due to additional attractive interactions with the ligand
at a greater distance from the reaction center, which
offset the deleterious effects in close. Another possibility
is that hydrophobic binding effects, which will certainly
be greater for adamantyl than for tert-butyl, override the
(
angle bending) could account for the tert-butyl group’s larger size
apparent.
(13) Manoharan, M.; Eliel, E. L. Tetrahedron Lett. 1984, 25, 3267.
14) The author has deposited atomic coordinates for this structure
(
with the Cambridge Crystallographic Data Centre. The coordinates
can be obtained, on request, from the Director, Cambridge Crystal-
lographic Data Centre, 12 Union Road, Cambridge, CB2 1EZ, UK.
(11) Vanhessche, K.; Sharpless, K. B. Chem. Eur. J ., in press.