Stereoselection in Intramolecular Diels-Alder Reactions of 2-Cyano-1-azadienes: Indolizidine and Quinolizidine Synthesis

Article abstract of DOI:10.1021/acs.orglett.6b00881

Progress toward understanding the scope and diastereoselectivity of intramolecular Diels-Alder reactions using 2-cyano-1-azadienes is described herein. The resulting cyanoenamine products are underutilized intermediates in organic synthesis. Assembly of the Diels-Alder precursors was achieved using an improved imine condensation/oxidative cyanation protocol. By this method, several highly substituted indolizidine and quinolizidine architectures were constructed. Quantum mechanical DFT calculations at the B3LYP/6-31+G(d) level of theory were performed for these cyclizations and provide insights into the origins of the observed diastereoselectivities.

Full text of DOI:10.1021/acs.orglett.6b00881

Letter
pubs.acs.org/OrgLett
Stereoselection in Intramolecular Diels−Alder Reactions of 2‑Cyano-
1-azadienes: Indolizidine and Quinolizidine Synthesis
Gidget C. Tay,† Nicholas Sizemore,*,‡ and Scott D. Rychnovsky*
Department of Chemistry, 1102 Natural Sciences II, University of California, Irvine, California 92697-2025, United States
S
* Supporting Information
ABSTRACT: Progress toward understanding the scope and
diastereoselectivity of intramolecular Diels−Alder reactions
using 2-cyano-1-azadienes is described herein. The resulting
cyanoenamine products are underutilized intermediates in
organic synthesis. Assembly of the Diels−Alder precursors
was achieved using an improved imine condensation/oxidative
cyanation protocol. By this method, several highly substituted
indolizidine and quinolizidine architectures were constructed.
Quantum mechanical DFT calculations at the B3LYP/6-
31+G(d) level of theory were performed for these cyclizations and provide insights into the origins of the observed
diastereoselectivities.
ndolizidines and quinolizidines are ubiquitous motifs found in
I
several familiesofalkaloidnaturalproducts.1 Bothringsystems
are present in many of the lipid-soluble toxins of poison dart frogs
isolated by the Daly group, including the indolizidines
gephyrotoxin and 261C (Figure 1).2 Many lycopodium alkaloids,
Figure 2. Indolizidines from IMDA reactions of 2-cyano-1-azadienes.
Figure 1. Examples of natural products containing indolizidine or
quinolizidine structural units (highlighted in blue).
in good yield.9 The ambiguous nature of reactivity for the
resulting α-cyano enamines products and their potential for
further elaboration caught our attention. These limited examples
wereapromisingstartingpointforcomplexalkaloidsynthesis, but
the scope of the IMDA reaction was underdeveloped. Herein, we
presentasystematicexplorationoftheIMDAreactionof2-cyano-
1-azadienes with unactivated alkenes.
Grierson and Fowler’s preparation of the 2-cyano-1-azadienes
from triflate activation of the corresponding amides9 proved to be
a challenging protocol to reproduce. In contrast, the one-pot
preparation recently outlined by Zhu and Masson appeared to be
promising.8 Their procedure combined an unbranched primary
amine, an unsaturated aldehyde, a cyanide source, and an oxidant.
In our hands, their method, which used IBX as the oxidant,
worked well with unbranched primary amines but gave low yields
with branched primary amines. A general method for the
synthesis of 2-cyano-1-azadienes was necessary for this IMDA
such as himeradine A, alsocontain quinolizidine substructures.3 A
variety of annulation strategies have been used to prepare these
interesting heterocycles.4 An intramolecular Diels−Alder
(IMDA) approach to the synthesis of highly functionalized
indolizidines, such as 261C, was an underdeveloped area in terms
of precursor access and stereochemical outcomes.5 The following
reportdescribes asimple, generalmethodforthepreparationof2-
cyano-1-azadiene precursors, as well as synthetic and computa-
tional studies regarding diastereoselectivity of cyclization.
Early investigations of 2-cyano-1-azadienes and related IMDA
reactions were described by the collaborative research of Fowler
and Grierson.6 One fascinating observation from these studies
was the competency of unactivated alkene dienophiles in such
IMDA reactions, which was attributed to the lowering of both
LUMO and HOMO energies by the 2-cyano group.7 Two
examples of the IMDA reaction using unactivated alkenes are
shown in Figure 2. Zhu and Masson reported on a high-
temperature reaction to provide an excellent yield of diaster-
eomers 2 and 3 as a 1:1 mixture.8 Grierson and Fowler found a 3:2
mixture of indolizidine diastereomers at a lower temperature, also
Received: March 26, 2016
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.6b00881
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
study. Zhu and Masson’s protocol was used as a starting point for
the development of an improved procedure.
Table 2. Optimization of the Diels−Alder Reaction with
Cyanoazadiene 9e
The one-pot method of Zhu and Masson relies on the rapid
condensation of unhindered primary amines with aldehydes.10
Branched primary amines, such as 8, were much less reactive and
competitive oxidation led to side reactions. Separating the
oxidation step allowed α-aminonitrile formation to be optimized.
Combining cinnamaldehyde 7 and branched amine 8 followed by
treatment of TMSCN led to complete conversion to the α-
aminonitrile product in quantitative yield after 2 h, as monitored
by NMR spectroscopy. Subsequent addition of IBX produced
reasonable yields of the 2-cyano-1-azadienes 9 in yields ranging
from 42 to 76% (Table 1). Recovered aldehyde was identified as a
d
entry temp (°C)
time
21 h
solvent
concn (M) yield (%)
a
c
c
c
c
c
c
c
1
2
3
4
5
6
7
8
180
180
220
150
170
170
170
170
wet toluene
wet toluene
wet toluene
wet toluene
wet toluene
wet toluene
wet toluene
dry toluene
0.04
0.04
0.04
0.04
0.04
0.04
0.02
64
a
a
a
a
b
b
b
29 h
15 min
24 h
23 h
24 h
24 h
24 h
61
30
no rxn
64
62
Table 1. Synthesis of 2-Cyano-1-azadienes by a Strecker
Reaction and an in Situ Oxidation
62
0.04
48
a
b
c
Heating with microwave. Heating thermally. The toluene was
d
stored over water. Yields were determined by 1H NMR spectroscopy
with respect to mesitylene as internal standard.
Scheme 1. Intramolecular Diels−Alder Reactions of 2-Cyano-
a
a
b
1-azadienes under Thermal Conditions
IBX yield
(%)
t-BuOCl yield
entry
R1
R2
R3
n
product
(%)
1
2
3
4
5
H
H
H
1
1
1
1
2
9a
9b
9c
9d
9e
42
76
47
54
52
86
74
67
75
80
Me
Me
H
H
H
Me
H
H
Me
H
H
H
a
Prestirred MgSO4, 7, and 8 in toluene at rt for 12 h, then added
TMSCN and MeOH. After 1 h, added TBAB and IBX, stirred for 16 h.
b
Combined 7 and 8 in toluene at rt for 40 min, then added TMSCN at
−78 °C. After 2 h, added Et3N and t-BuOCl.
side product, indicating that α-aminonitrile hydrolysis was an
issue. In considering other oxidants, we were attracted to tert-
butylhypochlorite, whichhadbeenusedforimineoxidationinthe
past.11 Freshly prepared tert-butyl hypochlorite reliably oxidized
the intermediate α-aminonitrile with an average yield of 76%
(Table 1). The optimized protocol is a convergent, one-pot
method and is effective for a variety of substrates. The new
oxidation procedure delivered the five 2-cyano-1-azadienes 9a−e
for IMDA studies.
The cyclization reaction was initially optimized using the
simple terminal alkene substrate 9e (Table 2). 1H NMR
spectroscopic analysis indicated the product was thermally stable
at 180 °C and that prolonged heating did not lead to significant
product decomposition (entries 1 and 2). Higher temperatures
ledtorapid decomposition (entry3). Noreactionwasobservedat
150 °C (entry 4), consistent with Zhu and Masson’s results but
not with Grierson and Fowler’s report (Figure 2). Reaction yields
were unaffected by the source of heating or reaction
concentration (entries 5−7). Solvents of varying polarity were
examined, andtoluenewasdeterminedtobethesolventofchoice.
Trace amounts of water helped to promote the Diels−Alder
cyclization compared to the use of anhydrous solvents (entries 6
and 8).12 A variety of additives were also investigated, but the
highest reproducible yield was achieved with no additives and the
diastereomeric ratio of products (18:19 = 1:2) remained
unaffected. A full account of the solvents and additives analyzed
is included in the Supporting Information.
a
Values in parentheses are products yields calculated with respect to
an internal mesitylene standard.
Substrates were selected to examine the effect of alkene
substitution on diastereoselectivity and to determine the viability
of sterically demanding cyclizations. The products were isolated
as diastereomeric mixtures, with the exception of 10 and 11,
which could be separated by flash chromatography. Stereo-
chemical configurations were assigned by 2D NMR methods and
extensive NOE analyses. The isolated yields ranged from 51% to
75%, demonstrating that the IMDA reaction is efficient across a
variety of substrates. The sterically demanding cyclizations of 9c
and 9d are particularly noteworthy as they represent the first
With a method for the synthesis of substrates and optimal
cyclization conditions identified, we were in a position to extend
the reaction to the other 2-cyano-1-azadienes (Scheme 1).
B
DOI: 10.1021/acs.orglett.6b00881
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
examples of using such a Diels−Alder reaction to generate
indolizidines bearing angular substituents or quaternary centers.
A major impetus of this project was to determine the
stereoselectivity of 2-cyano-1-azadiene IMDA reactions (Scheme
1). In each case, the major and minor products of the reaction
were the result of endo- and exo-IMDA reactions: there was no
observed isomerization of alkene or diene geometries in the
products. For the indolizidine products, the endo pathway was
favored for half of the cases and the exo pathway was favored for
the others, indicating the pathways are energetically similar. E-
Dienophile substrate 9b led to the highest dr (1:4) and a reversal
ofselectivity compared to Figure2 infavor ofthe exo product. 1,1-
Disubstituted substrate 9c, as well as the single quinolizidine case
(9e), also favored exo pathways (dr = 1:2). Although modest
diastereoselectivities were observed, more selective reactions
could be developed if the factors governing stereoselectivity were
understood. As such, we turned our efforts toward the
computational study of these IMDA reactions using the B3LYP
functional, which has been shown to accurately describe
stereoselection in related cycloaddition reactions.13
Detailed analysis for the computational study of the cyclization
of9bfollowsduetoitsdirectcorrelation toacomplexindolizidine
of interest (261C). It is noteworthy to mention that the endo:exo
selectivity is a reversal of previously known cases (Figure 2), and
the stereochemical relationships of 261C are present in minor
product (12). Additional computed structures (with truncated
phenethyl side chains) for all reactions in Scheme 1 as well as
explanations of computational methods appear in the Supporting
Information.
Each cyclization can proceed through four distinct diastereo-
meric transition states (TSs) (Figure 3). Cyclization from the
diene’s top face leads to the experimentally observed trans
relationship between the phenyl substituent on the diene and the
alkyl substituent on the tether (12′ and 13′), whereas the
unobserved products arising from a bottom face approach would
giverisetotheunobservedcisdisposition. Eachapproachcanhave
the tether oriented endo (12′ and 20) or exo (13′ and 21).
Computation correctly predicts trans approaches as being the
relevant reaction pathways in all cases, which can be rationalized
as aminimization of allylic strain and gauche interactions between
thetether, sidechain, andnitrilemoietiesinthetransitionstates.14
Computed TS structures corresponding to the formation of
observed products (TS-12′ and TS-13′, respectively) reveal
modest levels of asynchronicity, with C−N bond formation
advanced of C−C bond formation in both instances. The TS
leading to the major product (TS-13′) is the more asynchronous
of the two, with a short C−N distance (1.99 vs 2.09 Å for TS-12′)
and a longer C−C distance (2.36 vs 2.26 Å for TS-12′). The
energetic preference of the more asynchronous TS can be
rationalized as the one that most easily accommodates increased
steric bulk of the methyl group on the dienophile terminus.
Given the modest selectivity of these cyclizations and the
seemingly random product distribution from substrate to
substrate in Scheme 1, such comparisons of TS synchronicity
(vide supra) proved to be a valuable approach to evaluating the
data. The results of this analysis for all experimental cases are
shown in Table 3, where Δ corresponds to a measure of
synchronicity.15 Unsubstituted dienophiles (entries 1 and 5) are
sterically nondemanding substrates and prefer the more
synchronous TS geometry (i.e., the smaller Δ value). Conversely,
substituted dienophile substrates (entries 2−4) have a higher
steric demand and preferentially proceed through the more
asynchronous TS (larger Δ value). Interestingly, this approach
Figure 3. Examination of diastereoselection and TS structures for the
observed diastereomers in E-dienophile cyclization (analogous to
substrate 9b) calculated at the B3LYP/6-31+G(d) level. Forming
bonds are shown as dashed lines, and interatomic distances are given in
angstroms.
Table 3. Relationship of Calculated TS Distances (in Å) and
Observed Products As a Measure of Synchronicity
a
entry
1
substrate
product
C−N dist
C−C dist
Δ
9a
10
11
12
13
14
15
16
17
18
19
2.15
2.05
2.09
1.99
2.00
1.93
2.23
2.13
2.22
2.20
2.16
2.26
2.26
2.36
2.39
2.45
2.10
2.20
2.13
2.15
0.01
0.21
0.17
0.37
0.39
0.52
0.13
0.07
0.09
0.05
2
3
4
5
9b
9c
9d
9e
a
Absolute value of the difference between C−N and C−C distances.
provides consistent results regardless of tether length or the
placement of steric bulk on the dienophile, despite the nontrivial
reversals of observed endo/exo selectivities.
A comparison of experimental free energy differences (derived
fromproductdistributions)andcomputedfreeenergydifferences
(from DFT TS calculations) is provided in Figure 4. In all
examined cases, there was a computed energetic preference for
exo products corresponding to a positive value for ΔΔG⧧. The
overall difference between the experimental and computational
distributions leads to the ostensive overestimation of exo TS
energies by ∼0.8 kcal mol−1 for the indolizidine substrates. If one
C
DOI: 10.1021/acs.orglett.6b00881
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
ACKNOWLEDGMENTS
■
We thank Samantha Levine (UC Irvine) for early studies inamine
preparation. We thank Vertex Pharmaceuticals and the Allergan
Foundation (N.S.) for their generous financial support. We thank
UCIrvine, King’sCollege, andtheUniversityofScrantonfortheir
use of computational equipment, as well as Drs. Nathan Crawford
(UC Irvine), Fernando Clemente (Gaussian), and Christopher
Baumann (University of Scranton) for helpful discussions.
REFERENCES
■
(1) (a) Michael, J. P. Nat. Prod. Rep. 2007, 24, 191−222. (b) Michael, J.
P. Nat. Prod. Rep. 2008, 25, 139−165.
(2) (a) Daly, J. W.; Witkop, B.; Tokuyama, T.;Nishikawa, T.; Karle, I. L.
Helv. Chim. Acta 1977, 60, 1128−1140. (b) Daly, J. W.; Kaneko, T.;
Spande, T. F.; Garraffo, H. M.; Yeh, H. J. C.; Andriamaharavo, N. R.;
Andriantsiferana, M. Heterocycles 2003, 59, 745−757.
(3) Morita, H.; Hirasawa, Y.; Kobayashi, J. J. Org. Chem. 2003, 68,
4563−4566.
(4) (a) Mitchenson, A.; Nadin, A. J. Chem. Soc., Perkin Trans. 1 2000,
2862−2892. (b) Boger, D. L. Tetrahedron 1983, 39, 2869−2939.
(c) Boger, D. L. Chem. Rev. 1986, 86, 781−793. (d) Behforouz, M.;
Ahmadian, M. Tetrahedron 2000, 56, 5259−5288. (e) Buonora, P.;
Olsen, J.-C.; Oh, T. Tetrahedron 2001, 57, 6099−6138. (f) Memeo, M.
G.; Quadrelli, P. Chem.−Eur. J. 2012, 18, 12554−12582.
(5) Monbaliu, J.-C. M.; Masschelein, K. G. R.; Stevens, C. V. Chem. Soc.
Rev. 2011, 40, 4708−4739.
(6) (a) Teng, M.;Fowler, F. W. Tetrahedron Lett. 1989, 30, 2481−2484.
(b) Sisti, N. J.; Fowler, F. W.; Grierson, D. S. Synlett 1991, 1991, 816−
818. (c) Motorina, I. A.;Fowler, F. W.;Grierson, D. S. J. Org. Chem. 1997,
62, 2098−2105. (d) Motorina, I. A.; Grierson, D. S. Tetrahedron Lett.
1999, 40, 7211−7214. (e) Motorina, I. A.; Grierson, D. S. Tetrahedron
Lett. 1999, 40, 7215−7218.
(7) Sisti, N. J.; Motorina, I. A.; Tran Huu Dau, M.-E.; Riche, C.; Fowler,
F. W.; Grierson, D. S. J. Org. Chem. 1996, 61, 3715−3728.
(8) Fontaine, P.; Chiaroni, A.; Masson, G.; Zhu, J. Org. Lett. 2008, 10,
1509−1512.
(9) Sisti, N. J.; Zeller, E.; Grierson, D. S.; Fowler, F. W. J. Org. Chem.
1997, 62, 2093−2097.
(10) Chang, C.-F.; Li, C.-F.; Tsai, C.-C.; Chuang, T.-H. Org. Lett. 2016,
18, 638−641.
Figure 4. Plot of experimental and calculated ΔΔG⧧ (in kcal mol−1) for
aza-IMDA cyclizations and linear fit of indolizindine substrates.
corrects for the exo bias in the calculations, the correlation
between computed and experimental selectivities is very good.
Energy calculations using alternative density functionals and
Houk’s distortion−interaction models may provide more
accurate distributions and will be the subject of future studies.
In summation, a simplified procedure for the oxidation of
Strecker condensations provided access to IMDA precursors.
Subsequent cyclization of these precursors afforded indolizidine
and quinolizidine products in synthetically useful yields and
modest diastereoselectivities. These cyclizations are effective in
highly substituted examples including the formation of
quaternary centers. Computational studies have provided some
insight into the diastereoselectivity of these transformations.
These findings provide a foundation for further development of
such IMDA reactions and predictions of product distributions.
(11) (a) Padwa, A.; Gasdaska, J. R.; Hoffmanns, G.; Rebello, H. J. Org.
Chem. 1987, 52, 1027−1035. (b) De Kimpe, N.; Stevens, C. J. Org. Chem.
1993, 58, 2904−2906.
(12) Initial experiments were carried out with rigorous exclusion of air
and water. During optimization of the reaction, an increase in yield was
observed using wet toluene (see the Supporting Information for details).
Although the optimization studies were carried out on 9e, the improved
yieldsusingwettoluenewerealsoobservedempiricallyinthecyclizations
of 9a−d.
(13) (a) Wiest, O.; Montiel, D. C.; Houk, K. N. J. Phys. Chem. A 1997,
101, 8378−8388. (b) Tantillo, D. J.; Houk, K. N.; Jung, M. E. J. Org.
Chem. 2001, 66, 1938−1940. (c) Cayzer, T. N.; Paddon-Row, M. N.;
Moran, D.; Payne, A. D.; Sherburn, M. S.; Turner, P. J. Org. Chem. 2005,
70, 5561−5570.
(14)(a)Lilly, M.J.;Miller, N. A.;Edwards, A.J.;Willis, A.C.;Turner, P.;
Paddon-Row, M. N.; Sherburn, M. S. Chem.−Eur. J. 2005, 11, 2525−
2536. (b) Paddon-Row, M. N.; Moran, D.; Jones, G. A.; Sherburn, M. S. J.
Org. Chem. 2005, 70, 10841−10853. (c) Pearson, E. L.; Willis, A. C.;
Sherburn, M. S.; Paddon-Row, M. N. Org. Biomol. Chem. 2008, 6, 513−
522. (d) Paddon-Row, M. N.; Longshaw, A. I.; Willis, A. C.; Sherburn, M.
S. Chem.−Asian J. 2009, 4, 126−134.
ASSOCIATED CONTENT
* Supporting Information
TheSupportingInformationisavailablefreeofchargeontheACS
Publications website at DOI: 10.1021/acs.orglett.6b00881.
■
S
Computational methods, associated references, calculated
geometries and energies, and free energy diagrams (PDF)
Experimental procedures and characterization data for all
new compounds as well as stereochemical assignments for
IMDA products 10−19 and reaction optimization tables
(PDF)
AUTHOR INFORMATION
Corresponding Authors
■
*E-mail: nicholas.sizemore@scranton.edu.
*E-mail: srychnov@uci.edu.
Present Addresses
†Department of Chemical and Biological Sciences, Rockford
University, Starr Science Building, 5050 East State Street,
Rockford, IL 61108.
(15) The differences in C−C and C−N bond distances (1.54 and 1.47
Å, respectively) were not taken into account in Table 3 because the
presentation becomes complicated. Table SI1 in the Supporting
Information presents the Δ values, but corrected for the difference in
C−C and C−N bond lengths. The conclusions drawn from the analysis
using corrected bond lengths are unchanged.
‡Department of Chemistry, University of Scranton, 235 Loyola
Science Center, Scranton, PA 18510-4626.
Notes
The authors declare no competing financial interest.
D
DOI: 10.1021/acs.orglett.6b00881
Org. Lett. XXXX, XXX, XXX−XXX