Intramolecular Pyridinium Oxide Cycloadditions: Systematic Study of Substitution, Diastereoselectivity, and Regioselectivity

Article abstract of DOI:10.1002/chem.202100115

Intramolecular pyridinium oxide cycloadditions form complex polycyclic nitrogenous architectures. The diastereoselectivity and regioselectivity of pyridinium oxide cycloadditions was systematically investigated for the first time using complex substrates.

Full text of DOI:10.1002/chem.202100115

Communication
doi.org/10.1002/chem.202100115
Chemistry—A European Journal
&
Organic Chemistry |Hot Paper|
Intramolecular Pyridinium Oxide Cycloadditions: Systematic Study
of Substitution, Diastereoselectivity, and Regioselectivity
Yi Lu, Patrick N. Dey, and Christopher M. Beaudry*[a]
reogenic carbons bearing nitrogen.[6] In particular, all three car-
Abstract: Intramolecular pyridinium oxide cycloadditions
bons bearing the nitrogen atom in the himgaline structure are
form complex polycyclic nitrogenous architectures. The
stereogenic, making this target notably complex. Many syn-
diastereoselectivity and regioselectivity of pyridinium
thetic strategies have appeared to access nitrogenous bicycles,
oxide cycloadditions was systematically investigated for
and particularly successful approaches create multiple rings in
the first time using complex substrates. Predictably high
a single transformation with control of stereochemical out-
levels of diastereoselectivity and regioselectivity are ob-
comes.[7]
served, which can be attributed to minimization of steric
Cycloaddition reactions are particularly well suited for the
(syn-pentane) and torsional strain in the products. The re-
challenge of preparing multiple rings with control of stereo-
action is reversible under the reaction conditions, and it is
chemistry, and we wondered if such polycyclic nitrogenous
stereospecific with respect to the dipolarophile geometry.
frameworks could be prepared using a suitable cycloaddition.
We decided to focus on the tricyclic motif found in himgaline,
complete with the equatorial methyl substituent (5, Scheme 1).
Polycyclic nitrogenous architectures are widespread in biologi-
cally active natural products. For example, the tropane alka-
loids, exemplified by cocaine, are characterized by a 8-aza-bicy-
clo[3.2.1]octane structure, 1.[1] Amaryllidaceae alkaloids such as
siculine display a topologically distinct 1-azabicyclo[3.2.1]oc-
tane core (2).[2] Homopumiliotoxin 223G is a quinolizidine alka-
loid (3) isolated from poison dart frogs.[3] Finally, the intriguing
structure of the galbulimima alkaloid, himgaline, displays all
of these component aza-bicyclic fragments in a 2-azatricy-
clo[4.4.1.02,7]undecane architecture (4) (Figure 1).[4,5]
Any method capable of preparing this tricyclic architecture
could, at least in principle, be used to construct any of the
component bicyclic structures (1–3).
Such bridged polycyclic alkaloids represent enduring chal-
lenges for synthetic chemists. A notable difficulty associated
with the synthesis of the natural products shown are the ste-
Scheme 1. Retrosynthetic analysis of azatricyclo[4.4.1.02,7]undecanes.
We envisioned tricyclic molecules such as 5 arising from a
molecule such as enone 6, which contains functional handles
suitable for the construction of additional CÀC bonds. Consid-
eration of the topology of 6 suggested that an intramolecular
dipolar cycloaddition of a pyridinium oxide with a tethered
alkene dipolarophile (7) would give the required molecular
connectivity. Moreover, it appeared that the diastereoselectivity
of the cycloaddition may be controlled by a favored orienta-
tion of the molecular tether, which positions the methyl sub-
stituent in an equatorial position. The starting material for this
cycloaddition would be the relatively simple pyridinium oxide
8. One question was the regiochemistry of the cycloaddition;
specifically, regioisomeric tricycle 9 would arise from cycloaddi-
tion of conformer 10, and it was unclear to us which pathway
would be favored.
Figure 1. Polycyclic architectures featuring nitrogen atoms.
[a] Y. Lu, P. N. Dey, Prof. C. M. Beaudry
Department of Chemistry, Oregon State University
153 Gilbert Hall, Corvallis, OR 97333 (USA)
E-mail: christopher.beaudry@oregonstate.edu
Pyridinium oxide cycloadditions are well known in the litera-
ture. Since the first report by Katrizky in 1974,[8] there have
been dozens of intermolecular examples.[9] Intermolecular pyri-
Supporting information and the ORCID identification number(s) for the
author(s) of this article can be found under:
https://doi.org/10.1002/chem.202100115.
Chem. Eur. J. 2021, 27, 4028 –4032
4028
ꢀ 2021 Wiley-VCH GmbH

Communication
doi.org/10.1002/chem.202100115
Chemistry—A European Journal
dinium oxide cycloadditions require electron-poor alkene dipo-
larophiles (e.g. acrylonitrile, maleamide) for successful reactivi-
ty. Conversely, intramolecular pyridinium oxide cycloadditions
are much fewer in number (Scheme 2). Simple N-pentenyl pyri-
dinium oxide 11 reacts to form 12.[10] Substrate 13 contained a
substituted tether linking the dipole and dipolarophile, and it
cyclized to give 14.[11] In this case, the methyl substituents on
the phenyl ring were carefully chosen for their buttressing
effect, and less substituted congeners did not cyclize. Pyridini-
um oxide 15 featured a dipolarophile tethered to the pyridini-
um ring, rather than the nitrogen atom, and it gave the topo-
logically distinct bicycle 16.[12] It is notable that these intramo-
lecular examples do not require a polarized dipolarophile, and
simple terminal alkenes participate in the reaction.
(19) with 3-hydroxypyridines. Characterization of the product
was accomplished by NMR methods, and it appeared that the
product was the zwitterionic species 20 (not the 3-hydroxyl
pyridinium 21), as protonation of 20 to give 21 was observed
by NMR spectroscopy at pH < 0.[14] Pyridinium oxide 20 was
prone to decomposition over a few hours, and we commonly
advanced this material into the cycloaddition immediately
upon isolation (Scheme 3).
Scheme 3. Preparation of pyridinium oxides.
The intramolecular cycloaddition of 20 was investigated
using a variety of reaction conditions (Table 1). Simple heating
of 20 did not induce the intramolecular cycloaddition (entries 1
and 2). Heating of 20 in the presence of Bronsted acid
(entry 3) or Lewis acids (entry 4) were similarly ineffective at
promoting the reaction. Treatment of 20 with strong base re-
sulted in decomposition (entry 5). However, intramolecular cy-
cloaddition of 20 did occur under mildly basic conditions (en-
tries 6 and 7).[14] A survey of different bases identified that car-
bonate bases gave high yields of the cycloadduct 22 (entry 8).
Sequestration of the potassium counter ion did not improve
the reaction yield (entry 9). The reaction was largely insensitive
to the carbonate counter-cation, and Ag2CO3 was similarly ef-
fective in promoting the reaction (entry 10). Polar solvents
such as acetonitrile and butyronitrile gave the best reaction
Table 1. Identification of cycloaddition conditions.
Scheme 2. Intramolecular pyridinium oxide cycloadditions.
Gin published the only example of an intramolecular cyclo-
addition where stereochemistry in the tether led to good dia-
stereofacial discrimination of the dipolarophile (17!18).[13] The
electronics of the dipolarophile were carefully optimized to
promote the cycloaddition. The reaction was found to be re-
versible, which may be a result of the highly polarized dipolar-
ophile.
Clearly, such reactions have the ability to produce complex
polycyclic architectures; however, there has been no systemat-
ic investigation of substitution in such systems. Apart from
these relatively isolated examples shown in Scheme 2, no
study of the intramolecular pyridinium oxide cycloaddition has
been reported. Specifically, the tolerance of the reaction to
substitution, and the stereoselectivity of the reaction remain
unstudied.
We began our investigations by preparing pyridinium oxide
starting materials. There are multiple synthetic strategies for
preparing pyridinium oxides; however, we found the most
convenient method was the substitution of alkyl mesylates
Chem. Eur. J. 2021, 27, 4028 –4032
www.chemeurj.org
4029
ꢀ 2021 Wiley-VCH GmbH

Communication
doi.org/10.1002/chem.202100115
Chemistry—A European Journal
rates, limited the formation of unwanted by products, and
gave good selectivity for the major product. The best yields
were obtained in butyronitrile at 1308C, and we settled on
these as our standard conditions (entry 11).
The cycloaddition reaction gave two diastereomers, which
were separated and independently characterized. The structure
of the major and minor diastereomers was confirmed by 2D
NMR methods. Key COSY and NOESY correlations are shown in
Table 1. The major product of the reaction was the anticipated
diastereomer (22), which positions the tether methyl substitu-
ent in an equatorial orientation. The minor diastereomer (23)
displayed the methyl substituent in an axial orientation.
We next evaluated additional substrates to investigate the
ability of the intramolecular cycloaddition to create other com-
plex bicyclic products. Substitution on the dipolarophile and
the pyridinium oxide dipole was well tolerated, and in nearly
all cases the yields and diastereomer ratios were synthetically
useful. Substrate 24, which contains substitution on the pyridi-
nium ring and a more polarized dipolarophile (the type re-
quired in intermolecular dipolar cycloadditions), also reacted
under our standard conditions to give 25. Starting material 26
also contains a methyl substituent on the pyridine and an un-
activated dipolarophile, and it reacted smoothly to give 27. In-
terestingly, the reaction rate of 24 was not substantially differ-
ent than 22 or 26, which feature unactivated dipolarophiles.
Halogen substituents are also tolerated on the pyridinium
oxide. Substrate 28 contains a bromine atom on the pyridini-
um ring, and it reacts to form 29 in good yield. Chlorinated
substrate 30 had a simple unactivated dipolarophile, and it re-
acted in good yield to form 31. Brominated compound 32 also
participated to give 33. Finally, cyclic alkenes also participated
as dipolarophiles, and 34 reacted to give polycyclic product
35. Unfortunately, cyclization of trisubstituted alkene dipolaro-
philes such as 36 were unsuccessful (vide infra) (Scheme 4).
Additional substrates were prepared to test the hypothesis
that the equatorial positioning of the tether methyl group was
the stereochemical control element of the cycloaddition. When
the methyl substituent was moved to different positions on
the tether, the major diastereomer again featured an equatorial
methyl group. Substrate 37 was prepared, and the cyclization
similarly gave a major product (38) that displayed the equatori-
al methyl group. Starting material 39 has two tether methyl
substituents, and the corresponding cycloaddition product 40
results from positioning of both methyl groups equatorial. The
reaction was significantly more diastereoselective, and the
product (40) was formed as a single detectable (NMR spectros-
copy, TLC) diastereomer. Diastereomeric substrate 41 under-
goes the cycloaddition reaction, but one methyl group must
be axial in the product. In the event, product 42 was formed
as a 3:1 mixture of diastereomers. Taken together, these prod-
uct ratios strongly support the observation that diastereoselec-
tivity can be controlled by the tether substituents. All major
products were separated and the clean isomer was character-
ized by 2D NMR methods as described above (Scheme 5).
Our original hypothesis was that the stereochemical out-
come of the reaction was a result of the reaction kinetics. Spe-
cifically, we anticipated that the transition state leading to the
Scheme 4. Investigation of the cycloaddition substitution tolerance.
Scheme 5. Investigation of the diastereoselectivity.
major product would be favored if the tether methyl was
equatorial. This follows from Katritzky’s analysis on the inter-
molecular reaction.[15] The reversibility in Gin’s cycloaddition
was attributed to the polarized vinyl sulfone dipolarophile.[12]
However, we began to suspect our reactions with relatively
non-polarized dipolarophiles were under thermodynamic con-
trol as well. We noticed that the product ratios were often
somewhat variable, and more forcing conditions (higher tem-
Chem. Eur. J. 2021, 27, 4028 –4032
www.chemeurj.org
4030
ꢀ 2021 Wiley-VCH GmbH

Communication
doi.org/10.1002/chem.202100115
Chemistry—A European Journal
peratures, longer reaction times) generally gave superior prod-
uct ratios.
To investigate the reversibility of the reaction, substrate 43
was selected because in early investigations we observed ste-
reoselectivities that seemed to vary with temperature. More-
over, the major product was a crystalline solid, which enabled
us to further secure its structure and stereochemistry by X-ray
crystallographic studies. In this case we also isolated and char-
acterized the minor product isomer and found it to be a dia-
stereomer/regioisomer. Presumably, the regiochemistry of the
minor compound is a result of avoiding syn-pentane interac-
tions between the chloride substituent and the axial methyl
group (vide infra).
Initially, we performed the cycloaddition of 43 at lower tem-
peratures and observed a poor product ratio (Table 2, entry 1).
However, at higher temperatures the selectivity increased to
9:1 (entry 2), and longer reaction times further increased the
ratio to 18:1 (entry 3). Finally, resubjection of pure 44 (entry 4)
or pure 45 (entry 5) to butyronitrile at 1308C for 36 hours led
to an identical product mixture (18:1) in high yield. This is
strong evidence that the cycloaddition is reversible and under
thermodynamic control.
Scheme 6. Consideration of regioselectivity; key interactions highlighted.
not observed. In this case, the minor product regiochemistry
avoids steric interaction between the chlorine atom and the
axial methyl group. Both unobserved isomers (48 and 49)
would suffer from syn-pentane interactions of the methyl and
chloride groups.
Table 2. Reversibility studies.
We also wondered if the reaction was stereospecific with re-
spect to dipolarophile substituents. We prepared geometrical
isomers 50 and 51 and subjected them to our standard condi-
tions (Scheme 7). E-Isomer 50 reacted to give product 52, and
Z-alkene isomer 51 gave 53; the reaction is stereospecific with
respect to the alkene substituents. This also suggests that al-
though the reaction is reversible, it is likely proceeds with con-
certed bond formation. We observed that the cycloaddition of
E-configured 50 was a smooth transformation that occurred
under our normal conditions. However, the reaction yield with
51 was low, which results from the production of numerous
byproducts. Analysis of crystal structure data for the related
product 44 and models of 53 suggested that the pyrrolidine
methyl substituent projects directly into the concave bicyclic
structure leading to severe steric interactions. Such interactions
would destabilize the transition state leading to 53, which in
turn, would explain why by products are formed competitively.
Furthermore, these steric interactions would also explain the
reluctance of trisubstituted alkenes such as 36 to participate in
the cycloaddition reaction.
With the knowledge that the reaction is under thermody-
namic control, we ruminated on the regioselectivity of the cy-
cloaddition (Scheme 6). In the case of the relatively unsubsti-
tuted starting material 20, the isolated products are diastereo-
mers 22 and 23, and regioisomeric compounds 46 and 47
were not detected. The regiochemical preference can be attrib-
uted to the minimization of torsional strain in the major prod-
uct. Specifically, eclipsing of the carbonyl and the enone b-CÀ
H bond with the adjacent methine CÀH bonds is more severe
in 46 than in 22.
The major regioisomer often results from minimization of
torsional strain, but it can be over-ridden by other factors.
Chloride-containing substrate 43 undergoes cycloaddition
again giving the expected major product 44, which is the dia-
stereomer with the equatorial methyl. The minor isolated prod-
uct 45 is the diastereomer that positions the methyl group
axial. However, in this case the minor product displays the op-
posite regiochemistry. Isomeric compounds 48 and 49 were
Scheme 7. Investigations of stereospecificity.
Chem. Eur. J. 2021, 27, 4028 –4032
www.chemeurj.org
4031
ꢀ 2021 Wiley-VCH GmbH

Communication
doi.org/10.1002/chem.202100115
Chemistry—A European Journal
In conclusion, we have systematically studied the intramo-
lecular cycloaddition of pyridinium oxides. A range of substitu-
tion is tolerated in the substrates, and the reaction occurs with
good chemical yields. The reaction is reversible, and it is ste-
reospecific with respect to dipolarophile stereochemistry. Dia-
stereoselectivities can be controlled by the substituents on the
intramolecular tether. Regioselectivities result from torsional or
steric interactions present in the products. Efforts to apply this
transformation in the synthesis of polycyclic alkaloid targets
such as himgaline are underway in our laboratory and will be
reported in due course.
Keywords: alkaloids · cycloaddition · pericyclic reaction ·
pyridinium oxide · stereochemistry
[1] D. O’Hagan, Nat. Prod. Rep. 2000, 17, 435–446.
[2] Z. Jin, Nat. Prod. Rep. 2009, 26, 363–381.
[3] J. W. Daly, J. Nat. Prod. 1998, 61, 162–172.
[4] L. N. Mander, R. H. Prager, M. Rasmussen, E. Ritchie, W. C. Taylor, Aust. J.
Chem. 1967, 20, 1705–1718.
[5] a) U. Shah, S. Chackalamannil, A. K. Ganguly, M. Chelliah, S. Kolotuchin,
A. Buevich, A. McPhail, J. Am. Chem. Soc. 2006, 128, 12654–12655;
b) D. A. Evans, D. J. Adams, J. Am. Chem. Soc. 2007, 129, 1048–1049.
[6] A. Hager, N. Vrielkink, D. Hager, J. Lefranc, D. Trauner, Nat. Prod. Rep.
2016, 33, 491–522.
[7] A. Padwa, Russ. Chem. Bull. 2016, 65, 2183–2194.
[8] N. Dennis, A. R. Katritzky, T. Matsuo, S. K. Parton, J. Chem. Soc. Perkin
Trans. 1 1974, 746–750.
[9] A. R. Katritzky, N. Dennis, Chem. Rev. 1989, 89, 827–861.
[10] a) P. G. Sammes, R. A. Watt, J. Chem. Soc. Chem. Commun. 1976, 367–
368; b) M. P. Smith, C. George, A. P. Kozikowski, Tetrahedron Lett. 1998,
39, 197–200.
Crystallography data
Deposition Number 2051319 contains the supplementary crystallo-
graphic data for this paper. These data are provided free of
charge by the joint Cambridge Crystallographic Data Centre and
Fachinformationszentrum Karlsruhe Access Structures service
www.ccdc.cam.ac.uk/structures.
[11] B. S. Orlek, P. G. Sammes, D. J. Weller, J. Chem. Soc. Chem. Commun.
1993, 1412–1413.
[12] S. M. Bromidge, D. A. Archer, P. G. Sammes, J. Chem. Soc. Perkin Trans. 1
1990, 353–359.
Acknowledgements
[13] a) K. M. Peese, D. Y. Gin, Chem. Eur. J. 2008, 14, 1654–1665; b) K. M.
Peese, D. Y. Gin, J. Am. Chem. Soc. 2006, 128, 8734–8735.
[14] Acid titration of 20 suggests it is the zwitterionic pyridinium oxide not
the 3-hydroxylpyridinium. See Supporting Information.
[15] N. Dennis, B. Ibrahim, A. R. Katritzky, J. Chem. Soc. Perkin Trans. 1 1976,
2307–2334.
We thank Lev N. Zakharov for assistance with the X-ray crystal
structure of 44. We acknowledge funding from NSF (CHE-
1956401) and Oregon State University.
Conflict of interest
Manuscript received: January 12, 2021
Accepted manuscript online: January 12, 2021
Version of record online: February 10, 2021
The authors declare no conflict of interest.
Chem. Eur. J. 2021, 27, 4028 –4032
www.chemeurj.org
4032
ꢀ 2021 Wiley-VCH GmbH