Synthesis of cis,cis,cis,cis-[5.5.5.4]-1-azafenestrane with discovery of an unexpected dyotropic rearrangement

Article abstract of DOI:10.1002/anie.200500367

(Figure Presented) An unprecedented skeletal reorganization initially hindered a concise route based on a tandem [4+2]/[3+2] nitroalkene cycloaddition for the synthesis of the strained title compound. Conditions to suppress the observed dyotropic rearrangement were developed, and X-ray crystallographic analysis of the BF₃ derivative (see picture) of the azafenestrane revealed significant planarization around the central carbon atom.

Full text of DOI:10.1002/anie.200500367

Communications
Strained Polycycles
Synthesis of cis,cis,cis,cis-[5.5.5.4]-1-
Azafenestrane with Discovery of an Unexpected
Dyotropic Rearrangement**
Scott E. Denmark* and Justin I. Montgomery
Fenestranes[1] [fenestra (Lat.), window] are a unique family of
compounds with four fused carbocycles that share a central
quaternary carbon atom which exhibits planarizing distortion
(for example, 1, Scheme 1).[2] By changing ring sizes and ring-
Scheme 1. cis,cis,cis,cis-[5.5.5.5]-Fenestrane (1) and cis,cis,cis,cis-[5.5.5.5]-
1-azafenestrane·BH3 (2).
fusion configurations, variable planarization of the central
carbon atom results, thus giving chemists a tool with which to
probe one of the cornerstone theories of organic chemistry.[3]
Unfortunately, unsubstituted fenestranes are low-molecular-
weight hydrocarbons, and therefore crystallization for single-
crystal analysis and quantification of the distortions to the
central carbon atom is difficult.[4] By substituting a nitrogen
[*] Prof. S. E. Denmark, J. I. Montgomery
Department of Chemistry
University of Illinois
Urbana, IL 61801 (USA)
Fax: (+1)217-333-3984
E-mail: denmark@scs.uiuc.edu
[**] We are grateful to the National Institutes of Health for generous
financial support (GM30938). J.I.M. thanks the Procter & Gamble
Company, Abbott Laboratories, and Johnson & Johnson for
graduate fellowships. We thank Scott R. Wilson and Teresa Prussak-
Wieckowska of the UIUC George L. Clark X-Ray Facility for collection
and interpretation of X-ray data.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
3732
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/anie.200500367
Angew. Chem. Int. Ed. 2005, 44, 3732 –3736

Angewandte
Chemie
atom for one of the external bridgehead carbons, new
opportunities for X-ray crystallographic analysis exist.
We recently reported the first synthesis of a 1-azafenes-
trane, cis,cis,cis,cis-[5.5.5.5]-1-azafenestrane·BH3 (2),[5] which
featured the tandem [4+2]/[3+2] cycloaddition[6] of nitro-
cyclopentene 3 as the key step. Although the tandem cyclo-
addition allows for rapid and stereoselective formation of the
required skeleton, problematic steps in the synthesis of
nitroalkene 3 led to a poor overall yield of azafenestrane 2
(18 steps, 0.02% overall yield). Furthermore, azafenestrane 2
displayed only modest planarizing distortions to the central
carbon atom.[7] Therefore, an efficient route to an even more
strained azafenestrane was sought, and the successful realiza-
tion of that goal is described herein.
Scheme 3. a) THF, 08C, 1 h, then PhSeBr, 08C!RT, 1 h; b) H2O2, THF,
08C!RT, 30 min (78% from 10; 9/12 2:1).
nate[14] with phenylselenyl bromide, provided a mixture of
nitroselenides. Upon oxidation, syn selenoxide elimination
gave the desired nitroalkene 9, along with its double-bond
isomer 12 (9/12, 2:1) in 78% combined yield from 1-nitro-
cyclopentene. All attempts at chromatographic separation of
the two nitroalkenes were unsuccessful. Ultimately, the most
efficient way to separate the two nitroalkenes was simply to
carry out the next step of the azafenestrane synthesis with the
mixture of compounds. Trimethylaluminum-promoted [4+2]
cycloaddition of trisubstituted nitroalkene 9 with various
vinyl ethers takes place in less than one hour at ꢀ788C and
leaves tetrasubstituted nitroalkene 12 unreacted.
Contraction of one of the pyrrolizidine five-membered
rings in azafenestrane 2 should lead to increased planarization
of the central carbon atom.[8] Retrosynthetic analysis for
cis,cis,cis,cis-[5.5.5.4]-1-azafenestrane (4) is presented in
Scheme 2. The azetidine ring in 4 should be formed through
Treatment of nitroalkene 9 (Scheme 4) with n-butyl vinyl
ether in the presence of trimethylaluminum was expected to
provide nitroso acetal 6 (Scheme 2, R = n-butyl), the product
Scheme 2. Retrosynthetic analysis for cis,cis,cis,cis-[5.5.5.4]-1-azafenes-
trane (4).
an irreversible intramolecular displacement of an activated
alcohol 5, which is derived from nitroso acetal 6 through
ꢀ
hydrogenolysis. This reduction involves two N O-bond
cleavages and a reductive amination to form the pyrrolidine
ring in a single operation. Nitroso acetal 6 is the direct product
of tandem inter-/intramolecular [4+2]/[3+2] cycloaddition of
nitroalkene 9 and a suitable vinyl ether. Presumably, vinyl
ether 8 would approach from the less hindered side of the
nitroalkene (opposite the tethered olefin) to give an anti
relationship between the hydrogen atoms at the ring-fusion
sites in nitronate 7. An exo-fold, [3+2] cycloaddition[9] with
the unactivated dipolarophile[10] would then set the remaining
two stereogenic centers (final ring fusion and central carbon
atom) in the required sense for the all-cis azafenestrane
Scheme 4. a) AlMe3, CH2Cl2, ꢀ788C, 30 min (48% 14, 36% 15);
b) K2CO3, toluene, reflux, 2 h; c) SiO2, room temperature; d) H2
(26 atm), Raney Ni, MeOH, 14 h (46% from 15). ORTEP-3 plot of 18
(30% thermal ellipsoids).
of tandem [4+2]/[3+2] cycloaddition. However, the major
product of the reaction was compound 14, which contains an
azabicyclononane ring system and two five-membered rings!
The minor [4+2] cycloadduct 15 derived from endo approach
of n-butyl vinyl ether from the same face as the tethered
dipolarophile was also isolated. The structure of aminal 14
was established from studies carried out with the minor
cycloadduct 15: Thermal intramolecular [3+2] cycloaddition
of nitronate 15 with its tethered dipolarophile in refluxing
toluene gave nitroso acetal 16. However, upon purification
with silica gel, a new product, 17, which is a diastereomer of
aminal 14, was formed. The structure of 17, and by analogy
skeleton. The required 5-butenyl-1-nitrocyclopentene
9
would be available from 1-nitrocyclopentene (10) through
1,4-addition followed by regeneration of the nitroalkene.
A route analogous to that reported for nitroalkene 3
involving nitroallylation[11] was first investigated for the
synthesis of 9. Although the desired target was obtained, a
very poor overall yield for the sequence prompted the
development of a more efficient synthesis (Scheme 3). Treat-
ment of 1-nitrocyclopentene (10)[12] with 3-butenylcyanozinc
cuprate 11,[13] followed by trapping of the resulting nitro-
Angew. Chem. Int. Ed. 2005, 44, 3732 –3736
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3733

Communications
that of its diastereomer 14, was confirmed by X-ray crystallo-
(Scheme 7) led to formation of nitroso acetal 21 with no
rearrangement under the reaction conditions, thus providing
support for the proposed hypothesis. The crude reaction
mixture contained, in a 2:1 ratio, nitroso acetal 21 (endo
graphic analysis[15] of hydrogenation product 18, in which the
ꢀ
only N O bond has been cleaved.
This novel rearrangement that converts nitroso acetal 16
ꢀ
ꢀ
into aminal 17 involves breaking both an N O bond and a C
ꢀ
ꢀ
C bond with formation of a C O bond and a C N bond
(Scheme 5). Analysis of the 3D structure of nitroso acetal 16
clearly shows the two breaking bonds in near-perfect anti-
periplanar alignment. The ensuing transposition, known as a
dyotropic rearrangement,[16] can take place readily to give the
rearranged product 17.[17]
Scheme 7. a) AlMe3, CH2Cl2, ꢀ788C, 1 h, purification on Al2O3 (67%);
b) H2 (1 atm), Raney Ni, MeOH, (94%); c) MeOH, 3 h, room tempera-
ture, (70%); d) H2 (26 atm), Raney Ni, MeOH, room temperature,
15 h, (98%); e) H2 (26 atm), Raney Ni, 10% H2O-saturated EtOAc in
EtOAc (0.25m), 20 h, (85%); f) PPh3, diisopropyl azodicarboxylate
(DIAD), CH2Cl2, 08C, 40 min, then BH3·THF, ꢀ788C!RT, 1 h (87%).
Scheme 5. Dyotropic rearrangement of nitroso acetal 16.
Although this unexpected dyotropic rearrangement is
mechanistically intriguing and potentially useful synthetically,
its occurrence in the formation of the major product under the
conditions employed for the tandem cycloaddition reaction
precludes formation of the desired amino alcohol 5. However,
from conformational analysis of nitroso acetal 19 (the product
of tandem [4+2]/[3+2] cycloaddition with approach of the
dienophile from the opposite side to that occupied by the
tethered dipolarophile, and the precursor to aminal 14), a
means to prevent the rearrangement became apparent
(Scheme 6). The 1,2-oxazine ring in nitroso acetal 19-ax
approach of the dienophile from the side opposite the
tethered dipolarophile) and the minor nitronate (endo
approach of the dienophile from the same side as the
dipolarophile), which does not undergo spontaneous [3+2]
cycloaddition. Purification of nitroso acetal 21 on silica gel did
lead to partial rearrangement to aminal 23; however,
chromatography on basic alumina provided 21 in 67% yield.
With nitroso acetal 21 in hand, we were poised to
complete the synthesis in short order. Unfortunately, standard
hydrogenolysis conditions (H2, Raney Ni, MeOH) provided
the reduced, rearranged product 22, with no sign of the
desired amino alcohol. An investigation of reaction condi-
tions uncovered the fact that protic solvents induced the
dyotropic rearrangement. In fact, simply stirring nitroso
acetal 21 in methanol at room temperature provided aminal
23 in good yield. Consequently, new hydrogenation conditions
had to be developed that would allow reduction of nitroso
acetal 21 without promoting the rearrangement. Attempts to
carry out the hydrogenation in nonprotic, dry solvents were
unsuccessful; however, rearrangement did not take place
under these conditions. In the end, it was discovered that by
adding a controlled amount of water in the form of water-
saturated ethyl acetate (10% in dry ethyl acetate) to the
reaction mixture, the desired hydrogenolysis took place to
give amino alcohol 5 while still suppressing the rearrange-
ment.
Scheme 6. Conformations of nitroso acetal 19.
adopts a chair conformation, which places the butyloxy group
in an (anomerically stabilized) axial orientation. This con-
formation exhibits the required stereoelectronic alignment of
the two migrating bonds (bold); thus the rearrangement takes
place readily to give aminal 14. However, in a second low-
energy conformation, depicted in structure 19-eq, a chair flip
has occurred, and the migrating bonds (bold) are no longer in
alignment. By favoring this conformation, the rearrangement
should be suppressed, and hydrogenolysis might provide the
desired tricyclic amino alcohol 5. Calculations[18] suggest that
the two conformers are very close in energy. A nitroso acetal
derived from a bulkier vinyl ether might favor the equatorial
conformer and consequently deter or prevent the undesired
dyotropic rearrangement.
At the outset of the synthesis, the formation of the
azetidine ring from amino alcohol 5 was believed to be the
most challenging step of the proposed route. However, simple
treatment of 5 under Mitsunobu[19] coupling conditions led to
formation of the desired azafenestrane, which was efficiently
isolated as its borane complex cis,cis,cis,cis-[5.5.5.4]-1-aza-
fenestrane·BH3 (24) in 87% yield.[20] The availability of 24
from 5-butenyl-1-nitrocyclopentene 9 in only three steps and
50% overall yield demonstrates the power of the tandem
In the event, tandem cycloaddition of tert-butyl vinyl ether
with nitroalkene
9
promoted by trimethylaluminum
www.angewandte.org
3734
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2005, 44, 3732 –3736

Angewandte
Chemie
nitroalkene cycloaddition reaction for rapidly building molec-
ular complexity.
Keywords: cycloaddition · dyotropic rearrangement ·
fenestranes · nitroalkenes · strained molecules
.
To quantify the planarizing distortions at the central
carbon atom in 24, X-ray crystallographic analysis is required.
Although the azafenestrane–borane adduct is a crystalline
solid, crystals suitable for X-ray diffraction could not be
obtained.[21] Fortunately, simple treatment of 24 with boron
trifluoride etherate promoted an exchange to give the
crystalline BF3 adduct 25 (Scheme 8). Cooling of a warm,
saturated solution of 25 in hexane to room temperature gave
crystals of a quality suitable for X-ray crystallographic
[1] V. Georgian, M. Saltzman, Tetrahedron Lett. 1972, 13, 4315.
[2] M. Thommen, R. Keese, Synlett 1997, 231.
[3] a) H. van ꢀt Hoff, Arch. Neerl. Sci. Exactes Nat. 1874, 9, 445;
b) Le Bel, Bull. Soc. Chim. Fr. 1874, 22, 337.
[4] Low-temperature crystallization could possibly be employed;
see: J. Benet-Buchholz, T. Haumann, R. Boese, Chem. Commun.
1998, 2003.
[5] S. E. Denmark, L. A. Kramps, J. I. Montgomery, Angew. Chem.
2002, 114, 4296; Angew. Chem. Int. Ed. 2002, 41, 4122.
[6] a) S. E. Denmark, A. Thorarensen, Chem. Rev. 1996, 96, 137;
b) S. E. Denmark, J. J. Cottell in The Chemistry of Heterocyclic
Compounds: Synthetic Applications of 1,3-Dipolar Cycloaddi-
tion Chemistry Toward Heterocycles and Natural Products (Eds.:
A. Padwa, W. H. Pearson), Wiley-Interscience, New York, 2002,
pp. 83 – 167.
[7] Central angles in 2: N-C-C 116.18, C-C-C 116.68.
[8] Calculated central angles for 4: N-C-C 117.98, C-C-C 123.68
(DFT B88-LYP). All calculations were performed with CAChe
WorkSystem Pro Version 6.1.10, Fujitsu Limited.
[9] S. E. Denmark, D. S. Middleton, J. Org. Chem. 1998, 63, 1604.
[10] S. E. Denmark, C. B. W. Senanayake, Tetrahedron 1996, 52,
11579.
[11] D. Seebach, G. Calderari, P. Knochel, Tetrahedron 1985, 41, 4861.
[12] E. J. Corey, H. Estreicher, J. Am. Chem. Soc. 1978, 100, 6294.
[13] P. Knochel, M. C. P. Yeh, S. C. Berk, J. Talbert, J. Org. Chem.
1988, 53, 2390.
Scheme 8. a) BF3·OEt2, room temperature. ORTEP-3 plot of 25
(form 1, 30% thermal spheres).
analysis that formed in an unambiguous[22] space group P21/
n.[23] The two most populated crystal forms in the disordered
model exhibit similar planarization as defined by the angles
around the central carbon atom; form 1: N1-C1-C7 119.8(7)8
and C4-C1-C10 120.7(8)8, form 2: N1-C1-C7 119.2(8)8 and
C4-C1-C10 121.2(10)8. The degree of distortion agrees well
with calculated values for the corresponding parent hydro-
carbon.[2] Ab initio DFT calculations predict that the strain
energy of azafenestrane 4 is 17.8-kcalmolꢀ1 higher than that
of the previously synthesized cis,cis,cis,cis-[5.5.5.5]-1-aza-
fenestrane.[24]
In conclusion, the synthesis of cis,cis,cis,cis-[5.5.5.4]-1-
azafenestrane·BH3 (24) was completed efficiently in five steps
and 26% overall yield from 1-nitrocyclopentene by using a
tandem [4+2]/[3+2] cycloaddition of a nitroalkene as the key
step. Along the way, an unprecedented dyotropic rearrange-
ment was discovered that converts nitroso acetals into
tetracyclic aminals. The rearrangement is controlled by the
conformation of the six-membered ring in the nitroso acetal
precursors. By utilizing a bulky vinyl ether and developing
new hydrogenation conditions, the rearrangement was sup-
pressed, thus allowing the synthesis of the desired azafenes-
trane. The [5.5.5.4]-1-azafenestrane was analyzed by X-ray
crystallography as its BF3 adduct to quantify the planarizing
distortion around the central carbon atom. Efforts toward
even more strained azafenestranes, as well as investigations
into the reported dyotropic rearrangement and its use in
synthesis, are currently underway.
[14] S. E. Denmark, L. R. Marcin, J. Org. Chem. 1993, 58, 3850.
[15] Monoclinic, P21/c, crystal (0.64 ꢁ 0.56 ꢁ 0.10 mm3) from hexane:
a = 11.087(4) ꢂ, b = 14.064(5) ꢂ, c = 9.473(4) ꢂ, b = 96.733(7)8,
V= 1466.9(10) ꢂ3, 1 = 1.220 Mgmꢀ3. Bruker SMART CCD data
2qmax = 50.748, Mo radiation, l = 0.71073 ꢂ, w-scan profiles,
193(2) K, reflections (11547 measured, 2691 independent,
2020 > 2s(I)), limits (ꢀ13 ꢁ h ꢁ 13, ꢀ16 ꢁ k ꢁ 16, ꢀ11 ꢁ l ꢁ 11),
corrected for L-p effects and absorption (integration, m =
0.084 mmꢀ1, transmission 0.992 > 0.941). Direct-methods solu-
tion (Bruker SHELXTL) and full-matrix least-squares refine-
ment on F2 (Bruker SHELXTL) by using 179 parameters against
2691 data points, observed R1 = 0.040, wR2 = 0.094, residual
range 0.18 to ꢀ0.18 eꢂꢀ3. CCDC 262117 contains the supple-
mentary crystallographic data for this paper. These data can be
obtained free of charge from the Cambridge Crystallographic
Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
[16] M. T. Reetz, Adv. Organomet. Chem. 1977, 16, 33.
[17] The role of silica in promoting the dyotropic rearrangement
remains obscure.
[18] Local minima for 19-ax and 19-eq (both pseudo chairs in
minimized form) are within 1 kcalmolꢀ1 (PM3).
[19] a) O. Mitsunobu, Synthesis 1981, 1; b) P. G. Sammes, S. Smith, J.
Chem. Soc. Chem. Commun. 1983, 682.
[20] Full experimental details for 24 and all other compounds are
available as Supporting Information.
[21] Adduct 24 was soluble in many organic solvents. Thin needles
could be obtained from hexane at low temperature, but multiple
attempts to collect X-ray diffraction data were unsuccessful.
Crystals could be obtained from other nonpolar solvents (e.g.
tetraalkyl silanes, fluorocarbons) at ambient temperature, but
again, only weak diffraction patterns were observed.
[22] A detailed analysis of the unusual X-ray crystal structure of 25 is
available in the Supporting Information.
[23] Monoclinic, P21/n, crystal (0.60 ꢁ 0.20 ꢁ 0.04 mm3) from hexane:
a = 6.257(3) ꢂ, b = 14.424(6) ꢂ, c = 12.596(5) ꢂ, b = 91.932(8)8,
V= 1136.0(9) ꢂ3, 1 = 1.351 Mgmꢀ3. Bruker SMART CCD data
2qmax = 50.608, Mo radiation, l = 0.71073 ꢂ, w-scan profiles,
Received: January 31, 2005
Published online: May 11, 2005
Angew. Chem. Int. Ed. 2005, 44, 3732 –3736
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
3735

Communications
193(2) K, reflections (12439 measured, 2055 independent, 676 >
2s(I)), limits (ꢀ7 ꢁ h ꢁ 7, ꢀ17 ꢁ k ꢁ 17, ꢀ15 ꢁ l ꢁ 15), corrected
for L-p effects and absorption (integration, m = 0.112 mmꢀ1
,
transmission 0.995 > 0.946). Direct-methods solution (Bruker
SHELXTL) and full-matrix least-squares refinement on F2
(Bruker SHELXTL) by using 196 parameters and 237 restraints
against 2050 data points, observed R1 = 0.076, wR2 = 0.240,
residual range 0.28 to ꢀ0.26 eꢂꢀ3. CCDC 262116 contains the
supplementary crystallographic data for this paper. These data
can be obtained free of charge from the Cambridge Crystallo-
graphic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
[24] This value represents the total energy difference between 4 and
cis,cis,cis,cis-[5.5.5.5]-1-azafenestrane (B88-LYP) with an adjust-
ment for removal of a CH2 group in 4 (energy of unstrained CH2
determined by the average difference in total energy upon
minimization of ethane, propane, and butane); see: a) T. Dudev,
C. Lim, J. Am. Chem. Soc. 1998, 120, 4450; b) R. D. Bach, O.
Dmitrenko, J. Org. Chem. 2002, 67, 3884.
3736
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 3732 –3736