Diastereoselective osmylation and hydroboration of β,γ-unsaturated n,n-diisopropylamides and acid-catalyzed conversion to δ-lactones

Article abstract of DOI:10.1021/jo990119u

The title reactions of β,γ-unsaturated N,N-diisopropylamides occur with useful diastereofacial selectivity. The major diol isomer from osmylation of alkenes 1, 10, 11, and 12 in the presence of TMEDA at -78 °C corresponds to the facial preference shown in transition state model 41 (R(z) = H), while the opposite preference for 42 is observed with the Z-alkene 13. (Table 1). Hydroboration with 9-BBN does not show this inversion of diastereofacial selectivity for the Z-alkene. All of the results in Table 2 correspond to the usual preference for a transition state such as 45. Acid- catalyzed lactonization of the alcohols obtained in Tables 1 and 2 can be carried out with overall retention of configuration to afford δ-lactones. Butenolide 5 was prepared with 90% ee from alcohol 2a via osmylation followed by acid-catalyzed lactonization to 3 and elimination using SOCl₂/pyridine.

Full text of DOI:10.1021/jo990119u

4790
J . Org. Chem. 1999, 64, 4790-4797
Dia ster eoselective Osm yla tion a n d Hyd r obor a tion of
â,γ-Un sa tu r a ted N,N-Diisop r op yla m id es a n d Acid -Ca ta lyzed
Con ver sion to δ-La cton es
E. Vedejs* and A. W. Kruger
Chemistry Department, University of Wisconsin, 53706
Received J anuary 22, 1999
The title reactions of â,γ-unsaturated N,N-diisopropylamides occur with useful diastereofacial
selectivity. The major diol isomer from osmylation of alkenes 1, 10, 11, and 12 in the presence of
TMEDA at -78 °C corresponds to the facial preference shown in transition state model 41 (RZ )
H), while the opposite preference for 42 is observed with the Z-alkene 13. (Table 1). Hydroboration
with 9-BBN does not show this inversion of diastereofacial selectivity for the Z-alkene. All of the
results in Table 2 correspond to the usual preference for a transition state such as 45. Acid-catalyzed
lactonization of the alcohols obtained in Tables 1 and 2 can be carried out with overall retention of
configuration to afford δ-lactones. Butenolide 5 was prepared with 90% ee from alcohol 2a via
osmylation followed by acid-catalyzed lactonization to 3 and elimination using SOCl2/pyridine.
Previous studies in our laboratory have shown that the
Resu lts
â,γ-unsaturated amide 1 can be obtained with a high
level of enantiomeric excess by asymmetric protonation
of the corresponding enolate.1 We were therefore inter-
ested in determining whether 1 can be converted into
potentially useful chiral products by relying on the
R-carbon to direct electrophilic addition reactions by
sterically demanding reagents. Osmium tetroxide hy-
droxylation is one possibility for olefin functionalization
where the bulky allylic diisopropyl carboxamide group
might control facial selectivity.2 If either diol diastere-
omer 2a or 2b can be obtained with high selectivity, then
subsequent lactonization to 3 or 4 followed by â-elimina-
tion should afford the chiral butenolide 5 (from 3) or
en t-5 (from 4). In principle, the analogous sequence
would be possible from other chiral â,γ-unsaturated
amines as an entry into butenolide target structures. To
test the feasibility of this approach, we began with an
investigation of the osmium tetroxide hydroxylation using
racemic 1 as a representative substrate. The initial goal
was to learn whether alkene hydroxylation is possible
with an acceptable level of diastereofacial control. We also
hoped to demonstrate stereospecific conversion from the
amide 1 to the butenolide product 5 via a simple acid-
catalyzed lactonization-elimination sequence. If this can
be done without loss of enantiomeric purity starting from
(R)-1, then the bulky N,N-disopropylcarboxamide sub-
stituent might be used in a dual role to promote the
deracemization of â,γ-unsaturated amides and then to
direct the subsequent introduction of versatile oxygen
functionality.
Osmylation of r a c-1 using the catalytic conditions of
Van Rheenen and Kelly et al.3 proceeded smoothly at 0
°C and gave a 5:1 mixture of diols. The diastereomers
2a and 2b were separated by chromatography and were
converted into diastereomeric lactones 3 and 4, respec-
tively, by heating in the presence of acid catalysts.
Similar results were obtained with camphorsulfonic acid
(CSA) in refluxing toluene or with refluxing 10% H2SO4.
The latter procedure proved to be effective with a number
of simpler substrates to be described shortly. The 10%
H2SO4 method was also successful with the minor diol
2b, although the lactone 4 was obtained in modest 60%
yield together with a byproduct derived from pinacol
rearrangement.4 No attempt was made to optimize the
lactonization in the minor diol series. However, in the
case of 2a , the 10% H2SO4 method gave 3 together with
a minor (uncharacterized) byproduct containing vinylic
signals in the 1H NMR spectrum. Late in the study, after
nearly all of the correlation work had been completed, it
was found that the lactonization proceeds best when
performed with 10% H2SO4 in refluxing aqueous dioxane.
This method gave crystalline 3 as the sole product
detected by 1H NMR upon removal of solvent (81%
(3) Van Rheenen, V.; Kelly, R. C.; Cha, D. Y. Tetrahedron Lett. 1976,
17, 1973.
(4) The structure
i was established for the minor product by
correlation with 29b via Swern oxidation to the corresponding ketone
ii. The latter was not identical with the pinacol product, but treatment
with K2CO3/MeOH resulted in a 1:1 mixture of i and ii. Thus, the
pinacol product differs in the configuration of the enolizable carbon,
as expected from the assigned configurations of 29b and 2a if the
pinacol rearrangement occurs with stereospecific hydride migration
to the backside of the protonated tertiary hydroxyl group.
(1) Vedejs, E.; Lee, N.; Sakata, S. T. J . Am. Chem. Soc. 1994, 116,
2175.
(2) (a) Cha, J . K.; Christ, W. J .; Kishi, Y. Tetrahedron Lett. 1983,
24, 3943. Christ, W. J .; Cha, J . K.; Kishi, Y. Tetrahedron Lett. 1983,
24, 3947. Cha, J . K.; Christ, W. J .; Kishi, Y. Tetrahedron 1984, 40,
2247. (b) Vedejs, E.; McClure C. K. J . Am. Chem. Soc. 1986, 108, 1094.
(c) Fleming, I.; Lawrence, N. J .; Sarkar, A. K.; Thomas, A. P. J . Chem.
Soc., Perkin Trans. 1 1992, 3303. (d) Stork, G.; Kahn, M. Tetrahedron
Lett. 1983, 24, 3951. (e) Brimacombe, J . S.; Hanna, R.; Kabir, A. K.
M. S.; Bennett, F.; Taylor, I. D. J . Chem. Soc., Perkin Trans. 1 1986,
815. DeNinno, M. P.; Danishefsky, S. J .; Schulte, G. J . Am. Chem. Soc.
1988, 110, 3925. (f) Evans, D. A.; Kaldor, S. W. J . Org. Chem. 1990,
55, 1698. Panek, J . S.; Cirillo, P. F. J . Am. Chem. Soc. 1990, 112, 4873.
Vedejs, E.; Dent, W. H., III, Kendall, J . T.; Oliver, P. A. J . Am. Chem.
Soc. 1996, 118, 3556.
10.1021/jo990119u CCC: $18.00 © 1999 American Chemical Society
Published on Web 06/02/1999

â,γ-Unsaturated N,N-Diisopropylamides
J . Org. Chem., Vol. 64, No. 13, 1999 4791
Sch em e 1
isomeric lactone 6 was also isolated (5%). The structure
of the minor lactone was assigned on the basis of the
presence of a characteristic olefinic proton signal at δ 5.55
ppm in the 1H NMR spectrum as well as the infrared
frequency of 1756 cm-1. A similar sequence was carried
out from r a c-2a to provide a reference sample of 5/en t-5
(81%). Analysis by HPLC on a chiral stationary phase
proved difficult, but conditions were found that gave
near-baseline resolution and that were shown to detect
5% of the enantiomeric lactone in an authentic mixture
as an inflection point in the major peak. No inflection
point was detected in the sample of 5 obtained starting
from (R)-1 with 96% ee, so the product 5 is formed with
>90% ee, suggesting that there is no loss of configuration
over the sequence of lactonization and elimination steps.
The same sequence was also tested on a small scale
starting with enantiomerically enriched minor diol dias-
tereomer 2b. However, attempted elimination from 4
gave a mixture of three lactones with en t-5 as the minor
component (3-5%). Assay of this material was problem-
atic because of the order of peak elution. Significant
racemization was detected, and the major enantiomer
was clearly en t-5 as expected from the stereochemistry
of 2b and 4. However, only a qualitative assay for
enantiomeric purity was feasible (>60% ee and <80%ee)
due to partial peak overlap. In view of the limited
availability of 2b, no attempt was made to optimize this
sequence or to determine the source of partial racemiza-
tion at the stage of en t-5 in the minor isomer series.
The relative stereochemistry of the crystalline lactone
3 (from the major diol 2a ) was established using X-ray
techniques (see Supporting Information). Because the
C-O bonds in 3 are cis with respect to the six-membered
ring, it is likely that the lactonization process occurs with
retention of diol stereochemistry from 2a . This would
implicate a precedented acid-catalyzed acyl transfer via
7 as the lactonization mechanism (path a)7 and would
argue against an alternative (path b) involving backside
displacement of the protonated alcohol 8 by the amide
carbonyl. The latter sequence should produce the cis-
fused lactone 9, but no evidence to suggest formation of
this diastereomer was detected in any of the experiments.
Furthermore, the absence of lactone crossover products
in the two lactonizations (2a to 3; 2b to 4) indicates that
there is no enolization at the R-carbon under the lacton-
ization conditions. Thus, formation of 3 from 2a takes
place with retention of configuration at each of the
asymmetric carbons. This outcome is consistent with
earlier precedents for acid-catalyzed conversion of γ-hy-
droxyamides into lactones in hydroxylic solvents.7
isolated yield). Thus, optimization was necessary to
control side reactions, but the acid-catalyzed lactonization
pathway starting from 2a was stereospecific under all of
the conditions examined.
The osmylation diastereoselectivity at 0 °C (5:1 2a :2b)
is sufficiently high to allow purification of the major diol
by crystallization. However, higher selectivity is impor-
tant for other substrates where isomer separation is
difficult (see below), so several other experimental pro-
cedures were evaluated. The best diastereoselectivity
(11:1 2a :2b; 82% isolated yield of 2a ) resulted when the
osmylation was performed using a TMEDA-accelerated
stoichiometric OsO4 procedure at -78 °C (Scheme 1).5 At
room temperature, the TMEDA/OsO4 reagent gave a
product ratio of ca. 3.1:1, nearly the same as that
obtained under the catalytic osmylation conditions (3.4:
1).
The TMEDA-accelerated osmylation was repeated with
enantiomerically enriched (R)-1 (96% ee), and the result-
ing major diol was subjected to the aqueous dioxane (H2-
SO4-catalyzed) lactonization conditions, followed by treat-
ment with SOCl2/pyridine at 0 °C to convert 3 into the
butenolide 56 (75% yield based on 2). A small amount of
The low-temperature TMEDA hydroxylation of 1 pro-
ceeds with useful diastereofacial selectivity under control
of the amide R-carbon. It was therefore of interest to
determine whether similar results would be seen with
other â,γ-unsaturated amides. The prospects were evalu-
ated with the substrates 10-13 (Table 1).1 As expected
from the structural similarity, 12 gave essentially the
same product ratios as obtained from 1. All of the other
osmylations proved to be more highly selective, and the
low-temperature TMEDA procedure was consistently
superior to the catalytic method. However, one of the
osmylations (Z-trisubtituted alkene 13) was found to
proceed with opposite facial selectivity by comparison
(5) (a) Tomioka, K.; Nakajuma, M.; Koga, K. J . Am. Chem. Soc. 1987,
109, 6213. (b) Oishi, T.; Iida, K-i.; Hirama, M. Tetrahedron Lett. 1993,
34, 3573. (c) Donohoe, T. J .; Moore P. R.; Waring, M. J .; Newcombe,
N. J . Tetrahedron Lett. 1997, 38, 5027. (c) Vedejs, E.; Galante, R. J .;
Goekjian, P. G. J . Am. Chem. Soc. 1998, 120, 3613.
(6) Aben, R. W.; Hofstraat, R.; Scheeren, J . W. Recl.; J . R. Neth.
Chem. Soc. 1981, 100, 355.
(7) Martin, R. B.; Hedrick, R.; Parcell, A. J . Org. Chem. 1964, 29,
158. Hauser, C. R.; Adams, T. C., J r. J . Org. Chem. 1977, 42, 3029.

4792 J . Org. Chem., Vol. 64, No. 13, 1999
Vedejs and Kruger
Ta ble 1. Osm yla tion of Acyclic â,γ-Un sa tu r a ted Am id es
Sch em e 2
with 1 or with the other acyclic entries. This conclusion
did not come easily because of difficulties encountered
due to similar chromatographic behavior for the isomers
in several cases. Fortunately, the products 17a and 17b
could be separated from each other, and from a minor
contaminant (16a ) resulting from the presence of 9% of
12 in the starting 13.1 On the other hand, the minor diol
diastereomers could not be obtained pure starting from
alkenes 10-12. The major diol 14a from 10 could be
separated from 14b, but the latter was not obtained pure
and the NMR signals did not differ sufficiently to assay
the mixture. Fortunately, assay was possible after con-
version of the initial mixture of 14a and 14b to the
separable acetonides 18a and 18b (Scheme 2). In the
other osmylations, only the major diol diastereomers
(15a ; 16a ) could be purified and product ratios were
estimated from NMR integral comparisons of character-
istic signals assigned to the minor diol diastereomers
(15b δ 5.00, 3.65, 2.85 ppm; 16b δ 3.9, 2.7 ppm). These
assignments were guided by comparing fractions where
the minor diastereomers had been enriched by partial
separation or by osmylation at higher temperatures.
In the course of attempts to prove diol relative stere-
ochemistry, the purified major isomer 14a + en t-14a
(obtained from racemic 10) was converted into the
diastereomeric esters 19 and 20 by reaction with (S)-
naproxen in the presence of DMAP and N-ethyl-N-
dimethylaminopropylcarbodiimide (EDCI). One diaste-
reomer, 19, was characterized and crystallized, and X-ray
structure determination established the stereochemistry.
This information defines the relative configuration of the
major diol from 10 as shown in 14a /en t-14a . The
structures of diols 17a and 17b (from 13) were also
defined by X-ray crystallography. In this case, the minor
diastereomer 17a could be crystallized without deriva-
tization, and the knowledge of its relative stereochem-
istry defined the major diastereomer as 17b. However,
direct evidence for the structures of 15a and 16a could
not be obtained. These assignments were made on the
basis of NOE effects (GOESY method; gradient enhanced
nuclear Overhauser effect spectroscopy)8 after acid-
catalyzed conversion to the corresponding lactones 21 and
22. For purposes of comparison, GOESY experiments
were also performed with the lactone 23, obtained from
17b, and self-consistent results were obtained.
The acid-catalyzed cyclizations from diols 15a , 16a ,
and 17b were somewhat slower than the corresponding
laconization of 2a to 3. Relatively forcing conditions were
necessary (hours at 100 °C, 10% sulfuric acid), and this
raised the same concerns about retention of hydroxyl
stereochemistry as in the lactonizations of 2a and 2b.
Fortunately, the availability of diol diastereomers 16a
and 17b that differ only in the configuration at the
tertiary hydroxyl-bearing carbon (different rotamers are
shown at the secondary carbon) simplifies the arguments.
(8) Stonehouse, J .; Adell, P.; Keeler, J .; Shaka, A. J . J . Am. Chem.
Soc. 1994, 116, 6037.

â,γ-Unsaturated N,N-Diisopropylamides
J . Org. Chem., Vol. 64, No. 13, 1999 4793
Sch em e 3
is in the allylic position, although a variety of interesting
methods for controlling diastereoselectivity are described
in this series.10b-d We conclude that the osmylation
reagent prefers to avoid the bulky allylic carboxamide
group. The results indicate that the favored geometry for
osmylation of 13 differs from that of the more constrained
analogue 24, presumably due to the conformational
freedom at the allylic C-C bond.
Several other comparisons were made that suggest a
dominant role for steric factors in the osmylations. The
N,N-dimethyl analogue of 1 was explored briefly under
the TMEDA conditions. Osmylation selectivity was lower
(7:1 at -78 °C), but the major product had the same
relative configuration, as shown by acid-catalyzed lac-
tonization to 4. Furthermore, osmylation of the corre-
sponding ester 2611 (Scheme 3) proceeded with almost
no selectivity (1.2:1 diastereomer ratio; major product not
assigned). Thus, the isopropyl groups are important for
obtaining a practical level of diastereoselectivity.
It was also of interest to compare the osmylation
results with hydroborations conducted with 9-BBN, a
sterically discriminating electrophile that reacts with a
well-precedented facial selectivity pattern.12 Product
configurations can easily be established via oxidative
workup followed by the same acid-catalyzed lactonization
procedure as used in the osmylation series. This system
has the advantage that both of the diastereomeric lac-
tones expected from the alkenes 1, 12, and 13 are already
described in the literature.13
The hydroborations proved to be relatively slow, and
heating was necessary (3-24 h in refluxing THF).
However, each of the three test substrates reacted with
a clear preference for one major alcohol diastereomer
(Table 2). The purified hydroxyamides 27, 28, and 29
were lactonized using 10% H2SO4 as before, and the
resulting 30, 31, and 32 were identified by comparisons
with literature NMR data.13 All three lactones have the
same relative stereochemistry. Thus, hydroboration oc-
curs with the same facial preference with the E- or the
Z-trisubstituted alkenes 12 and 13, in contrast to the
osmylations. The major diastereomers obtained with
9-BBN correspond to the osmylation facial preference
seen with the Z-alkene 13, but the hydroboration facial
preferences for alkenes 10, 11, and 12 are inverted
compared to those found for the corresponding osmyla-
tions.
Because 16a and 17b afford unique products 22 and 23,
reversible SN1 cleavage of the tertiary C-O bond before
or after cyclization can be ruled out. According to the
NOE relationships summarized in Scheme 2, both 22 and
23 retain the oxygen stereochemistry expected from cis-
hydroxylation of the starting E- or Z-alkenes. The con-
figuration of 17b is known from the X-ray structure
determination for 17a . Thus, conversion to 23 must take
place with retention of relative stereochemistry at both
oxygenated carbons. Therefore, the configuration at the
secondary hydroxyl of 16a can be deduced from the
stereochemistry of 22, assuming retention in the lacton-
ization step by analogy to the conversion from 17b to 23.
Finally, the relative stereochemistry at tertiary hydroxyl
in 15a can be assigned from the NOE results and from
the survival of tertiary hydroxyl configuration in the
related lactones 22 and 23 made under similar condi-
tions.
A comparison of the major diol products obtained from
the â,γ-unsaturated amides in Table 1 reveals that the
relative stereochemistry follows the same pattern with
respect to the asymmetric R-carbon in all cases except
for the Z-trisubstituted alkene example 13. A similar
inversion of osmylation facial preferences in Z- vs E-
alkenes has been observed in certain enoates2d,e and in
4-substituted 2-pentenes.2b However, the amide substitu-
ent differs in its potential for osmium complexation
compared to the examples explored previously. The
constrained Z-trisubstituted alkene 249 (Scheme 3) was
therefore studied to help determine whether the stere-
ochemical outcome in the unconstrained analogue 13
could involve directed osmylation mediated by the amide
carbonyl group. Internal complexation of osmium by the
amide carbonyl would be expected to form the syn
hydroxylation product 25b, a substance that is analogous
to 17b in relative configuration. However, the reaction
gave 25a as a single diastereomer, even at room tem-
perature, and the configuration of the secondary hydroxyl
was anti with respect to the amide according to the
vicinal coupling, J 1,2 ) 9.6 Hz. This result is consistent
with an earlier report of osmylation anti to an allylic
carboxamide group in a cyclopentene environment.10a We
could find no reports of specific directing effects involving
amide carbonyl in other studies where amide nitrogen
(10) (a) Palmer, C. F.; McCague, R.; Ruecroft, G.; Savage, S.; Taylor,
S. J . C.; Ries, C. Tetrahedron Lett. 1996, 37, 4601. (b) Osmylation syn
to allylic N-H is known in an amide example, but this selectivity may
be influenced by additional factors: King, S. B.; Ganem, B. J . Am.
Chem. Soc. 1991, 113, 5089 and references therein. For leading
references to syn direction by other allylic heteroatoms, see ref 5c. (c)
Krysan, D. J .; Rockway, T. W.; Haight, A. R. Tetrahedron: Asymmetry
1994, 5, 625 and references therein. (d) Reetz, M. T.; Strack, T. J .;
Mutulis, F.; Goddard, R. Tetrahedron Lett. 1996, 37, 9293. Paz, M.
M.; Sardina, F. J . J . Org. Chem. 1993, 58, 6990. Hauser, F. M.;
Ellenberger, S. R. J . Am. Chem. Soc. 1984, 106, 2458. Hauser, F. M.;
Rhee, R. P. J . Org. Chem. 1981, 46, 227.
(11) Chavan, S. P.; Zubaidha, P. K.; Ayyangar, N. R. Tetrahedron
Lett. 1992, 33, 4605.
(12) (a) Transition state proposal: Evans, D. A.; Bartroli, J .; Godel,
T. Tetrahedron Lett. 1982, 23, 4577. Still, W. C.; Barrish, J . C. J . Am.
Chem. Soc. 1983, 105, 2487. Houk, K. N.; Rondan, N. G.; Wu, Y.-D.;
Metz, J . T.; Paddon-Row: M. N. Tetrahedron 1984, 40, 2257. (b)
Examples: Schmid, G.; Fukuyama, T.; Akasaka, K.; Kishi, Y. J . Am.
Chem. Soc. 1979, 101, 259. Midland, M. M.; Kwon, Y. C. J . Am. Chem.
Soc. 1983, 105, 3725. Burgess, K.; Cassidy, J .; Ohlmeyer, M. J . J . Org.
Chem. 1991, 56, 1020. Burgess, K.; Ohlmeyer, M. J . J . Org. Chem.
1991, 56, 1027.
(13) (a) Petrzilka, M.; Felix, D.; Eschenmoser, A. Helv. Chim. Acta
1973, 56, 2950. (b) Forzato, C.; Nitti, P.; Pitacco, G. Tetrahedron
Asymmetry 1997, 8, 4101.
(9) Vedejs, E.; Sakata, S. Unpublished Results.

4794 J . Org. Chem., Vol. 64, No. 13, 1999
Vedejs and Kruger
Ta ble 2. Hyd r obor yla tion of â,γ-Un sa tu r a ted Am in es
w ith 9-BBN
F igu r e 2. Crystal structure of 17a .
4-center (osmaoxetane) mechanisms had not yet
emerged.14 Despite these uncertainties, the results sug-
gested that allylic SiMe2Ph or SO2Ph substituents control
facial selectivity due primarily to their steric bulk. In the
current study involving the osmylation of the trisubsti-
tuted analogues 12 and 13, the same selectivity trends
are apparent and a similar steric rationale is plausible
for 13 if the hindered N,N-diisopropylamide environment
can be accommodated within a transition state geometry
analogous to 36. The role of 34 is less clear and must be
reevaluated in light of recent mechanistic findings.
Differences due to the potential involvement of amide
carbonyl as a ligand that might coordinate to osmium
must also be considered. However, literature precedents10a
as well as the anti osmylation observed with the cyclic
alkene 24 argue against internal coordination by amide
carbonyl oxygen.
The issue of amide steric effects in the acyclic sub-
strates is potentially complicated, but crystal structure
evidence suggests a simplifying approximation as out-
lined below. The X-ray structure of diol 17a reveals an
interesting arrangement of staggered isopropyl substit-
uents (Figure 2). In one of the isopropyl groups, the C7,-
C6,C8 angle is bisected by the plane of the amide NCd
O subunit. The other isopropyl is turned so that the
C10,C9,C11 angle is bisected by the C6-H bond. This
arrangement places the C9-H near the C2-H and close
to the plane defined by the carbonyl carbon (C1) and the
C2-H bond, resulting in the effective separation of
methyl groups and the proximity of the C9 and C2
hydrogens (shown in bold font). While the structure 17a ′
may reflect crystal lattice effects or contributions by
hydrogen bonding, a search of the Cambridge crystal-
lographic database turned up three other examples of
R-branched N,N-diisopropylamides, all of which have a
similar local conformation.15 The proximity of the R-hy-
drogen (corresponding to C2-H) and one of the isopropyl
F igu r e 1. Major pathways for osmylation of 4-substituted
pent-2-enes.
Discu ssion
In a prior study, the inversion of osmylation diaste-
reofacial preferences from E-alkenes 33 to the Z-alkenes
35 was attributed to steric factors in the competing
transition states (Figure 1). Essentially identical selec-
tivities were found in the case of 35a and 35b despite
the expected difference in substituent donor/acceptor
properties, and the pathway corresponding to a proposed
transition state geometry 36 was favored by a 4:1 ratio
at room temperature.2b The isomeric E-alkenes 33a and
33b reacted with the inverted facial preference, corre-
sponding to a favored geometry such as 34. Key geo-
metrical details were left unspecified in the prior report
because a consensus for the 5-center rather than the
(14) (a) Corey, E. J .; Noe, M. C.; Grogan, M. J . Tetrahedron Lett.
1996, 37, 4899. (b) DelMonte, A. J .; Haller, J .; Houk, K. N.; Sharpless,
K. B.; Singleton, D. A.; Strassner, T.; Thomas, A. A. J . Am. Chem.
Soc. 1997, 119, 9907.
(15) For an example, see: Lutz, G. P.; Wallin, A. P.; Kerrick, S. T.;
Beak, P. J . Org. Chem. 1991, 56, 4938.

â,γ-Unsaturated N,N-Diisopropylamides
J . Org. Chem., Vol. 64, No. 13, 1999 4795
etries, depending on the extent of rehybridization and
the nature of substrate interactions with developing
bonds and osmium ligands. However, recent develop-
ments in the mechanistic interpretation of osmium
tetroxide reactions suggest that ground state preferences
can be quite important for the TMEDA-accelerated
reactions at low temperature. According to the mecha-
nistic proposal by Corey et al.,17 the stereochemistry-
determining step in diamine-accelerated reactions is
olefin complexation to produce a labile 20-electron com-
plex that undergoes subsequent rearrangement to an
osmate ester. Possible transition structures based on the
Corey geometry are shown in Figure 4. Structure 41
assumes relatively little carbon rehybridization in the
olefin subunit, corresponding to the π-interaction of the
OsO4-TMEDA complex with one of the local energy
minima for the alkene (structure 38′). Geometry 41
encounters no interactions between osmium ligands with
the allylic (C2) methyl or amide isopropyl substituents
and accounts for the facial preference seen in all of the
osmylations of Table 1 with the single exception of the
Z-alkene 13. Alternative transition structure 42 corre-
sponding to another alkene local energy minimum (39′
in Figure 3) also avoids isopropyl-osmium ligand inter-
actions if bonding occurs from below, but in this case
ligand interactions with the allylic methyl are possible.
Therefore, 41 is favored over 42 with RZ ) H. On the
other hand, 41 is destabilized when RZ ) CH3 and 42
becomes the preferred transition state. Analogous struc-
tures derived from the ground state rotamer 40/40′
(Figure 3) are unlikely because they would encounter
osmium ligand interactions with isopropyl groups on one
side of the olefin (complex formation from below) or
ligand interactions with the carbonyl oxygen and C2
methyl on the other side (complex formation from above).
In the case of the Z-alkene, there would also be a severe
interaction between RZ (C4 methyl) and the amide. Any
single one of these interactions can be reduced by turning
the isopropyls or the C1-C2 bond, but such conforma-
tional changes would result in increased isopropyl inter-
actions with other groups because the amide geometry
would have to deviate from the arrangements shown in
Figure 3.
F igu r e 3. Selected ground state geometries.
The catalytic osmylations at room temperature are not
likely to follow an early transition state model such as
that described above for the TMEDA-accelerated (sto-
ichiometric Os) reactions. Carbon isotope effect studies
indicate that the five-center transition state has signifi-
cant C-O bonding at both olefinic carbons.14 The catalytic
process therefore involves substantial rehybridization at
carbon, and there are differences in the arrangement and
number of osmium ligands relative to the TMEDA
reactions. However, the diastereofacial selectivity is
similar (Table 1) if the difference in reaction temperature
is taken into account. The consistent selectivity pattern
indicates that rehybridization at the olefinic carbons does
not strongly affect the relative energies of competing
transition states even though there must be some change
in transition state geometries due to bond angle changes
and the differences in ligand geometry.
hydrogens (correponding to C9-H) is characteristic, and
the amide geometries can be represented by dihedral
angle variations between the two limiting cases 37a and
37b (Figure 3). Within this range of geometries, the
osmium tetroxide amine complex can interact with an
adjacent alkene (R′ ) RERZC ) CR) without encountering
the isopropyl methyls while the latter avoid interactions
with the C2 substituents. Similar local conformers will
be used to represent the amide environment in subse-
quent drawings, as in the ground state structures 38-
40. These geometries represent the three alkene rotamers
around the allylic C2-C3 bond that have the olefinic Cd
C subunit eclipsed with one of the allylic bonds and
therefore correspond to the local energy minima.16
The favored transition state geometries for osmylation
need not be similar to the favored ground state geom-
A consistent selectivity pattern in early as well as
relatively advanced transition states is easy to under-
stand in the Z-alkene case. A comparison of the reactant-
(16) Allinger, N. L.; Hirsch, J . A.; Miller, M. A.; Tyminski, I. J . J .
Am. Chem. Soc. 1968, 90, 5773. Lowe, J . P. Prog. Phys. Org. Chem.
1968, 6, 1. Wiberg, K. B. J . Am. Chem. Soc. 1986, 108, 5817. Dorigo,
A. E.; Pratt, D. W.; Houk, K. N. J . Am. Chem. Soc. 1987, 109, 6591.
Bader, R. F. W.; Cheeseman, J . R.; Laidig, K. E.; Wiberg, K. B.;
Breneman, C. J . Am. Chem. Soc. 1990, 112, 6530.
(17) Corey, E. J .; Sarshar, S.; Azimioara, M. D.; Newbold, R. C.; Noe,
M. C. J . Am. Chem. Soc. 1996, 118, 7851.

4796 J . Org. Chem., Vol. 64, No. 13, 1999
Vedejs and Kruger
transition state than in the currently favored 5-center
mechanism. Depending on the extent of rehybridization,
43a will be somewhat destabilized by eclipsing effects.
If the C2 carbon is turned counterclockwise to increase
the degree of staggering as shown in 43b, then isopropyl
interactions with osmium ligands are increased unless
the amide geometry is restricted to resemble local
conformer 37b. An alternative clockwise rotation pro-
duces 43c, a geometry that would also increase stagger-
ing of bonds, but at the cost of C2-methyl interactions
with osmium ligands. Structure 43c has the C2 methyl
“inside” the developing bonds, a situation that is ener-
getically unfavorable according to recent calculations for
the osmylation of allylic alcohol analogues.18 Transition
state geometries between 43a and 43b are therefore
considered most likely for the E-alkene osmylations in
Table 1, depending on the extent of transition state
rehybridization.
No inversion of facial selectivity is seen in the hydrobo-
rations. All of the examples in Table 2 correspond to
favored transition state 45, an arrangement that follows
the well-known pattern with the bulky group anti to the
developing bonds in a staggered geometry with the
smallest group “inside”.12a Reagent bulk in the case of
45 is confined to an area near the substituents RE and
RZ, and only the compact B-H bond approaches the C2
methyl regardless of olefin geometry. In short, both the
osmylations and the hydroborations are controlled largely
by steric factors, but the results are different because the
reagents differ in shape and in the nature of the reacting
bonds.
Con clu sion
Osmylation or hydroboration of the â,γ-unsaturated
N,N-diisopropylamides (Table 1, Table 2) occurs with
useful diastereoselectivity. As observed in several earlier
osmylations,2b,d,e the Z-alkene 13 reacts with the opposite
diastereofacial preference compared to the E-isomer 12.
Acid-catalyzed lactonization of the alcohols described in
Tables 1 and 2 can be carried out with overall retention
of configuration to afford δ-lactones. Starting with the
alcohol 2a , the sequence provides access to the hydroxy
lactone 3 and elimination affords the butenolide 5 with
>90% enantiomer excess.
Exp er im en ta l Section
Gen er a l. Materials were purified as follows. TMEDA was
dried over CaH2 and distilled. CH2Cl2 was distilled from P2O5.
All other reagents were used as received. Osmylation sub-
strates 1,1 the N,N-dimethyl analogue of 1,1 10,19 11,1912,1 13,1
24,9 and 2620 were prepared according to literature procedures.
Gen er a l NMO-Med ia ted Osm yla tion P r oced u r e. The
procedure is a modification of the method of Van Rheenen.3
To a stirred solution of N-methylmorpholine N-oxide (1.5
equiv) and the alkene (1.0 equiv) in 10:1 acetone/H2O at (0 °C
or room temperature) was added OsO4 (0.02-0.05 equiv, 2.5%
w/v solution in t-BuOH, Aldrich). After TLC analysis indicated
consumption of the alkene, excess Na2HSO3/Na2S2O5 was
added and the darkened reaction was stirred for 30 min at
F igu r e 4. Proposed transition state geometries.
like 42 with the more advanced transition state 44
(monodentate amine; room-temperature osmylation) shows
dihedral angle changes along the C4-C3-C2-H seg-
ment, but the geometry remains near the expected local
energy minimum (nearly eclipsed in alkene-like 42; more
staggered in the partly rehybridized 44). There are other
subtle changes, but the geometry of 44 retains the
important features that favor 42 vs alternative geom-
etries as discussed earlier. A similar argument can be
made in the E-alkene series, but there are additional
geometries to consider. The partially rehybridized transi-
tion state geometry 43a corresponds closely to that of 34
in the earlier proposal that focused on minimal C2
substituent interactions with osmium ligands. These
factors would be more important in an osmaoxetane
(18) Haller, J .; Strassner, T.; Houk, K. N. J . Am. Chem. Soc. 1997,
119, 8031.
(19) Majewski, M.; Mpango, G. B.; Thomas, M. T.; Wu, A.; Snieckus,
V. J . Org. Chem. 1981, 46, 2029.
(20) Henin, F.; Mortezaei, R.; Muzart, J .; Pete, J .-P.; Piva, O.
Tetrahedron 1989, 45, 6171.

â,γ-Unsaturated N,N-Diisopropylamides
J . Org. Chem., Vol. 64, No. 13, 1999 4797
room temperature. To this solution was added ether, and the
solution was dried (MgSO4) and the solvent removed (aspira-
tor).
the structural assignment. Molecular ion calcd for C9H14O3
170.09430, found m/e ) 170.0957, error ) 8 ppm; IR (neat,
cm-1) 3508, O-H; 1759, CdO; 300 1H MHz NMR (CDCl3, ppm)
δ 3.97 (1H, dd, J ) 11.6, 5.2 Hz) 2.46 (1H, q, J ) 7.0 Hz) 2.01-
1.87 (4H, m) 1.69-1.62 (3H, m) 1.52-1.25 (2H, m) 1.20 (3H,
d, J ) 7.0 Hz).
Gen er a l TMEDA-Med ia ted Osm yla tion P r oced u r e.5 To
a stirred solution of the alkene and TMEDA (1.1-1.5 equiv)
at -78 °C was added OsO4 (solution in CH2Cl2), and the
solution turned dark. After stirring for 2 h the reaction was
tested for the presence of the starting alkene by TLC. Once
the alkene was consumed, the osmate ester was reduced with
1:1 saturated NaHSO3/THF (10 mL) (reflux, 2 h). Brine was
added, and the resulting solution was extracted with EtOAc
(3×). The combined organic extracts were then dried (MgSO4)
and evaporated (aspirator) to give the crude product.
Diols 2a (Ma jor ) a n d 2b (Min or ). The general NMO
procedure (rt) using 11 (213 mg, 0.90 mmol), NMO (126 mg,
1.1 mmol), and OsO4 (0.25 mL of a 2.5% v/v solution in t-BuOH,
0.02 mmol, Aldrich) afforded 254 mg (99%) of a 3.2:1 diaste-
reomer ratio (2a (major):2b (minor)) that was >90% diols. The
residue was purified by flash chromatography on EM silica
gel 60 (33 × 1.6 cm, 50 mL and then 8 mL fractions). Fractions
13-16 gave 52 mg of 2b (21%) and fractions 18-25 afforded
165 mg of 2a (68%); 4:1 CH2Cl2/ether eluent; analytical TLC
on EM silica gel 60, 4:1 CH2Cl2/ether, Rf ) 0.50 (2b, minor),
Rf ) 0.45 (2a , major). Ma jor d ia st er eom er 2a . r a c-
(2R,1′S,2′S)-2-(1′,2′-Dih yd r oxycycloh exyl)-N,N-d iisop r o-
p ylp r op a n a m id e: solid; molecular ion calcd for C15H29NO3
271.21470, found m/e ) 271.2146, error ) 0 ppm; IR (neat,
Enantiomerically enriched 3 was prepared by the same
sequence starting from (R)-1 of 96% ee and performing the
lactonization on 98 mg of enantiomerically enriched diol 2a
to give 50 mg (82%) of (3R,3aS,7aS)-3 after chromatography
as described above, oil, ([R]25 ) -46, c ) 0.05).
D
La cton iza tion of th e Min or Diol 2b. r a c-(3R,3a R,7a R)-
3a -Hyd r oxy-3-m eth ylh exa h yd r oben zofu r a n -2-on e (4). To
r a c-2b (25 mg, 0.09 mmol) was added 0.27 mL of a 10% H2-
SO4 solution (v/v) (5 equiv), and the mixture was heated to
reflux for 2 h. The acid layer was extracted with ether (3 × 2
mL). The combined organic layers were dried (MgSO4) and
evaporated (aspirator) to give 13 mg of an oil. 1H NMR analysis
indicated a 3.5:1 ratio of 4:2b with no sign of 3. An isomeric
product was also present (ca. 10%; Rf ) 0.47 in 1:1 EtOAc:
hexane), identified as the ketone derived from pinacol rear-
rangement of 2b. The structure of the pinacol contaminant
was confirmed by correlation with 29b (see Supporting Infor-
mation).4 The residue was purified by preparative layer
chromatography on EM silica gel 60 (20 × 20 × 0.02 cm), 4:1
CH2Cl2/ether eluent (Rf ) 0.36), which afforded 9 mg (60%) of
4 as an oil. Analytical TLC on EM silica gel 60, Rf ) 0.37.
Molecular ion calcd for C9H14O3 170.09430, found m/e )
170.0944, error ) 1 ppm; IR (neat, cm-1) 3482, O-H; 1757,
CdO; 300 MHz NMR (CDCl3, ppm) δ 4.11 (1H, dd, J ) 11.8,
4.8 Hz) 2.55 (1H, q, J ) 7.7 Hz) 2.04-1.25 (9H, m) 1.20 (3H,
d, J ) 7.7 Hz).
Elim in a tion of En a n tiom er ica lly En r ich ed 3 to 57
Usin g SOCl2/P yr id in e. To a solution of (3R,3aS,7aS)-3 (45
mg, 0.26 mmol) in 0.5 mL of pyridine at 0 °C was added thionyl
chloride (60 µL, 0.79 mmol). After 15 min the reaction was
warmed to room temperature and the solution was stirred for
30 min. Ether (5 mL) was added, and the resulting solution
was extracted with 1.5 N HCl (3 × 3 mL). The ether layer
was dried (MgSO4) and evaporated to an oil. The initial residue
was purified by flash chromatography (13 × 1 cm, 2 mL
fractions), 9:1 hexane:ether. Fractions 8-10 gave 2 mg of the
minor lactone 6. Fractions 13-17 gave 30 mg (75%) of pure
(S)-5 ([R]25D ) +54, c ) 0.03). Comparison of a racemic sample
of 57 and the enantiomerically enriched (S)-5 by HPLC on a
chiral stationary phase (93:3 hexane/EtOH, (R,R)-Whelk-O 1
(Regis), flow ) 1.0 mL/min, tR19.3 min (R), 20.1 min (S))
established that the ee was greater than 90% favoring the (S)
enantiomer. Baseline resolution could not be achieved, but an
authentic 95:5 mixture of enantiomers prepared from the
enriched and racemic samples showed a clear inflection point
for 90% ee that was absent in the enriched sample (>90% ee).
To establish the structure of 6, a similar experiment was
performed starting with racemic racemic 3 and r a c-6 was
obtained as an oil; analytical TLC on EM silica gel 60, 1:1
hexane/ether, Rf ) 0.59. Molecular ion calcd for C9H12O2
152.08370, found m/e ) 151.0758; M - 1, 151.0758; IR (neat,
cm-1) 1756, CdO; 300 MHz 1H NMR (CDCl3, ppm) δ 5.55 (1H,
br s) 4.86-4.76 (1H, m) 3.20-3.09 (1H, m) 2.36-2.27 (1H, m)
2.19-2.09 (2H, m) 2.00-1.89 (1H, m) 1.69-1.52 (1H, m) 1.47-
1.33 (1H, m) 1.31 (3H, d, J ) 6.9 Hz).
1
cm-1) 3442, O-H; 3236, O-H; 1604, CdO; 300 MHz H NMR
(CDCl3, ppm) δ 6.46 (1H, s) 4.08 (1H, sept, J ) 6.6 Hz) 3.60-
3.39 (2H, m) 2.82 (1H, q, J ) 7.2 Hz) 2.50 (1H, br s) 1.90-
1.42 (6H, m) 1.41-1.22 (14H, m) 1.18 (3H, d, J ) 7.0 Hz); 13
C
NMR (75 MHz, CDCl3, ppm) δ 177.5 s, 74.7 s, 73.4 d, 48.5 d,
45.9 d, 37.2 d, 30.8 t, 29.9 t, 21.2 t, 21.2 t, 20.9 q, 20.8 q, 20.2
q, 20.2 q, 12.7 q. Min or d ia ster eom er 2b. r a c-(2R,1′R,2′R)-
2-(1′,2′-Dih yd r oxycycloh exyl)-N,N-d iisop r op ylp r op a n a -
m id e: colorless oil; molecular ion calcd for
C15H29NO3
271.21470, found m/e ) 271.2155, error ) 3 ppm; IR (neat,
1
cm-1) 3419, O-H; 3263, O-H; 1604, CdO; 300 MHz H NMR
(CDCl3, ppm) δ 5.20 (1H, br s) 4.25-4.05 (1H, m) 3.80-3.40
(2H, m) 3.71 (1H, ddd, J ) 9.9, 4.4, 2.2 Hz) 2.91 (1H, q, J )
7.2 Hz) 1.79-1.32 (14H, m) 1.29-1.19 (9H, m); 13C NMR (75
MHz, CDCl3, ppm) δ 176.7 s, 74.4 s, 70.2 d, 49.2 d, 46.3 d,
35.5 d, 29.8 t, 23.1 t, 21.3 t, 21.1 q, 20.6 q, 20.5 q, 14.5 q.
P r ep a r a tion of 2a Usin g TMEDA/OsO4. To a solution of
11 (272 mg, 1.15 mmol) and TMEDA (208 µL, 1.38 mmol) in
2.0 mL of CH2Cl2 at -78 °C was added OsO4 (350 mg, 1.38
mmol) in CH2Cl2 (2 × 1.0 mL). After 2 h, TLC analysis
indicated 1 had been consumed. The solution was allowed to
warm to room temperature, and the solvent was evaporated
to produce a brown residue. After the addition of 10 mL of
THF, 1.6 g of NaHSO3, and 0.75 mL of water, the solution
was heated to reflux for 2 h. The solution was then diluted
with ether (30 mL), dried (MgSO4), and evaporated to a green
oil (314 mg). 1H NMR assay of this residue indicated a 11:1
diastereomer ratio and >95% conversion to diols. The residue
was purified by flash chromatography on EM silica gel 60 (23
× 1.6 cm, 7 mL fractions; 4/1, CH2Cl2/ether eluent). Fraction
11 gave 11 mg of a mixture of 2a and 2b. Fractions 12-33
gave 253 mg (82%) of pure 2a .
Lacton ization of Diol 2a. r a c-(3R,3aS,7aS)-3a-Hydr oxy-
3-m eth ylh exa h yd r oben zofu r a n -2-on e (3) a n d (3R,3a S,-
7a S)-3. To r a c-2a (200 mg, 0.74 mmol) in 1.8 mL of dioxane
was added 1.6 mL of a 10% H2SO4 solution (v/v) (4 equiv), and
the solution was heated to reflux for 9 h. The acid layer was
extracted with ether (3 × 10 mL). The combined organic layers
were dried (MgSO4) and evaporated (aspirator) to give 126 mg
of a white solid. 1H NMR analysis indicated >90% conversion
to product. Crystallization of the initial product from ether
afforded 77 mg of 3 (62%). The mother liquor (44 mg) was
purified by flash chromatography on EM silica gel 60 (11 × 1
cm, 4:1 CH2Cl2/ether eluent, 3 mL fractions, 25 mL prerun).
Fractions 4-7 gave 24 mg of 3 (19%). The materials were
combined to afford 101 mg (81%) as a white crystalline solid.
Pure r a c-3 was obtained by crystallization from ether/
hexane, mp 120.0-121.0 °C. X-ray crystallography confirmed
Ack n ow led gm en t. This work was supported by the
National Institutes of Health (GM44724).
Su p p or tin g In for m a tion Ava ila ble: X-ray data tables
for 3, 17a , and 19. NMR spectra. Procedures and characteriza-
tion for 14a , 18a ,b 19, 15-17, 21-23, and 25-29, and
correlations with 30-32. This material is available free of
charge via the Internet at http://pubs.acs.org.
J O990119U