Synthesis of Diverse N-Acryloyl Azetidines and Evaluation of Their Enhanced Thiol Reactivities

Article abstract of DOI:10.1021/acs.orglett.7b00788

Acyl azetidines exhibit nonplanar hybridization, leading to lower amide-like character of the corresponding (O)C-N bonds. This impacts N-acryloyl azetidines by producing enhanced electrophilicy at appended Michael acceptors. Herein, reactivity data are reported in the presence of glutathione (GSH) in phosphate buffer (pH 7.4) at 37 °C. Wide reactivity ranges are observed by varying substitution at the Michael acceptor or by modulating the electron-withdrawing character of substituents at the C3 position of the azetidine.

Full text of DOI:10.1021/acs.orglett.7b00788

Letter
pubs.acs.org/OrgLett
Synthesis of Diverse N‑Acryloyl Azetidines and Evaluation of Their
Enhanced Thiol Reactivities
Maximilian D. Palkowitz, Bo Tan, Haitao Hu, Kenneth Roth, and Renato A. Bauer*
Lilly Research Laboratories, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, Indiana 46285, United States
S
* Supporting Information
ABSTRACT: Acyl azetidines exhibit nonplanar hybridization,
leading to lower amide-like character of the corresponding
(O)C−N bonds. This impacts N-acryloyl azetidines by
producing enhanced electrophilicy at appended Michael
acceptors. Herein, reactivity data are reported in the presence
of glutathione (GSH) in phosphate buffer (pH 7.4) at 37 °C.
Wide reactivity ranges are observed by varying substitution at
the Michael acceptor or by modulating the electron-with-
drawing character of substituents at the C3 position of the
azetidine.
zetidines serve as structural motifs in natural products,1 as
2
Ascaffolds in diversity-oriented synthesis, and as synthetic
intermediates that can be reacted with nucleophiles in ring-
opening reactions.3 They are also widely used in medicinal
chemistry and can be found in approved drugs.4 Azetidines
have recently emerged as scaffolds to present Michael acceptors
to nucleophilic residues in proteins associated with disease (e.g.,
ARS-853, Figure 1a).5 The most commonly used Michael
acceptor in covalent drug discovery at the present time is the
acrylamide (e.g., ibrutinib, Figure 1a).6 Despite the widespread
use of azetidines and acrylamides in drug design, the reactivity
of acrylamides bound at the nitrogen of azetidine is not well
understood. Indeed, N-acryloyl azetidines may suffer from
attrition during lead generation due to enhanced and/or poorly
understood reactivity. A recent report by the Szostak group7
and earlier crystallographic work by Ohwada8 propose that acyl
azetidines are geometrically distorted in a manner that leads to
lower amide-like character of the (O)C−N bond. Woodward
previously used the height, h, of a theoretical pyramid formed
by the three carbon atoms of an amide, to assess and compare
nonplanarity among amides in the context of penems (Figure
1b).9 Comparison of h values of a reactive penem9 with
azetidine-N-toluamide and piperidine p-chlorobenzamide, from
published crystallography data,8 suggested to us that the amides
of N-acylazetidines exhibit degrees of planarity between
Figure 1. (a) Structures of ibrutinib and ARS-853. (b) Pyramid height
conventional amides and the β-lactam class of amides that
used as a model to evaluate nonplanarity in amides.
enjoy various grades of reactivity (Figure 1). A departure from
amide planarity may impact acryloyl azetidines by producing
enamides (e.g., polymerizability: methacryloyl aziridine >
azetidine > pyrrolidine > piperidine > dimethylamine).10
Pfizer,11 Amgen,12 and others13,14 have elegantly profiled the
reactivity of diverse acrylamides against glutathione (GSH), an
enhanced electrophilicity at the β-acrylamide carbon relative to
acrylamides appended to common ring sizes such as 5-, 6-, and
7-membered rings. In the area of polymer science, Suzuki et al.
have observed that the yields and rates of methacrylamide
polymerization correlate well with the decreasing ring size of
appended amides, providing a key example of how amide
nonplanarity can influence the reactivity of corresponding
Received: March 16, 2017
© XXXX American Chemical Society
A
DOI: 10.1021/acs.orglett.7b00788
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
abundant biological nucleophile. Unfortunately, while thiol
reactivity data of acrylamides appended to common rings have
been reported, data on N-acryloyl azetidines are notably
absent.15 To effectively design covalent inhibitors around such
azetidine scaffolds, we set out to explore and report on their
reactivity.
which possess acrylamides appended to 5-, 6-, and 7-membered
rings, also reacted more slowly than 1, confirming that the
distorted amide of 1 may be inducing enhanced thiol reactivity
as hypothesized. Interestingly, data obtained from the reactions
of 4−7, whose structures vary six bonds away from the reacting
acrylamide β-carbon, suggested that distal functionalities can
have significant effects on reactivity via through-bond electron-
withdrawing effects (t1/2’s ranged from 5.90 to 33 h).
Homopiperazine 8 was more reactive than 4−7 (although
not as reactive as azetidine 1), perhaps due to the electronic
effect of a protonated distal nitrogen under the reaction
conditions (t1/2 = 3.58 h).
Azetidine 1 was targeted as a simple synthetic target to
evaluate reactivity (Scheme 1). A 3-benzyloxy substituent was
Scheme 1. Effect of Ring Size on the Reactivity of Appended
a
Acrylamides
To further assess the effect of distal functional groups,
various derivatives of azetidine 1 were also synthesized and
tested (Scheme 2). All products were conveniently derived
Scheme 2. t1/2 Values for the Reaction with GSH in
a
Parentheses
a
Half-life (t1/2) measurements for ibrutinib were used as a control for
assay reproducibility. Reactions were monitored by MS at 0.1−1 mM
electrophile, 10 mM glutathione, 70 mM phosphate buffer (pH = 7.4),
b
c
and 30% CH3CN at 37 °C. Reference 11. NR2 = 3-(4-
phenoxyphenyl)-1H-pyrazolo[3,4-d]pyrimidin-4-amine (cf. Figure 1a).
a
Reactions were monitored by MS at 0.1−1 mM electrophile, 10 mM
glutathione, 70 mM phosphate buffer (pH = 7.4), and 30% CH3CN at
37 °C.
incorporated to avoid volatility and to provide a chromophore
for UV-guided purification. The structure of 1 also allowed for
the subsequent introduction of a range of design changes for
evaluating structure−reactivity relationships (vida infra). Using
established conditions for assessing reactivity with GSH,11,12
mass spectrometry-guided reaction monitoring was used to
compare azetidine 1 with the open-chain congener 2, ring
analogs 3−8, and the approved drug ibrutinib (Scheme 1). To
confirm that thiol addition fully accounted for the consumption
of Michael acceptors, both starting material disappearance and
GSH-adduct formation were monitored over time. Products
derived from mechanisms other than GSH-addition were not
detected, and comparisons of the rate of GSH addition to the
rate of substrate disappearance supported thiol addition as the
exclusive reaction pathway. In the assay, azetidine 1 exhibited a
half-life (t1/2) of 2.78 h under the reaction conditions (10 mM
GSH, 70 mM phosphate buffer, pH 7.4, 37 °C), whereas the
ring-opened analog 2 reacted nearly 10-fold more slowly with a
t1/2 of 21.4 h. Importantly, the saturated propionamide analog
of 1 was found to be unreactive (not shown), suggesting that
GSH was reacting with 1 at the Michael acceptor and not
through azetidine ring opening. Analogs 3−8, and ibrutinib,
from commercially available building blocks with final
introduction of electrophiles achieved through treatment of
azetidine hydrochloride salts with 1.1 equiv of acryloyl chloride
and 2 equiv of triethylamine at −78 °C in CH2Cl2 (vida infra,
Scheme 4). Moving the oxygen of 1 a single atom further from
the acrylamide provided ether 9 that exhibited a longer half-life
in our GSH assay (t1/2 = 3.62 h). Replacing the oxygen atom
with a methylene reduced electrophilicity in 10 as expected
(t1/2 = 6.50 h). Interestingly, moving the benzyl group of 1
from an ether linkage to a carbinol linkage afforded a slight
increase in reactivity (11, t1/2 = 1.81 h). Since electronic effects
are not significant in this structural modification (1 vs 11),
conformational effects of the benzyl group may be responsible
for the change in reactivity. Molecular modeling suggests, for
example, that in polar aqueous environments the benzyl of 11
can π-stack with the acrylamide, resulting in a possible increase
in amide pyramidylization and increased reactivity of 11.16
Nitrogen-substituted azetidines 12 and 13 exhibited lower and
higher thiol reactivities than 1, respectively, based on the
substituents on the nitrogen linker (tertiary amine 12, t1/2
=
3.36 h; sulfonamide 13 t1/2 = 0.72 h). Azetidine 14 is embedded
B
DOI: 10.1021/acs.orglett.7b00788
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
one C−O bond connecting the ether of 1 to its azetidine.18 In
the case of trans-2-methyl azetidine 15, the resolved (+)- and
(−)-enantiomers were found to be equally reactive to GSH.
Next, to fully understand the range of potential reactivities
accessible in azetidine scaffolds, substitution effects at
acrylamide group were evaluated (Scheme 3). Methyl
substitution at the Michael acceptor, exemplified by azetidines
16−19, prevented thiol addition within the reaction time scale
at the temperature analyzed (t1/2 > 60 h). However, 2-chloro
substitution in 20 significantly increased reactivity relative to 1
(t1/2 = 0.178 h v 2.78 h). The 2-fluoro analog 21 was prepared
and isolated, but was not sufficiently stable on standing or for
analysis in the GSH assay. Interestingly, 3-chloroacrylamide 22
afforded no net change in the rate of GSH addition relative to
1, likely as a result of a competition between the reactivity-
enhancing and -suppressing characteristics of the 3-chloride
(i.e., increased electronegativity vs increased steric bulk).
Dehydroalanine 23 afforded an azetidine of diminished
reactivity (t1/2 = 10.7 h), even lower than the reactivity of the
approved electrophilic drug ibrutinib (t1/2 = 7.36 h).
Propynamides were also synthesized and tested; terminal-
and chloro-alkynes 24 and 25 were extremely reactive with t1/2
values less than 5 min, whereas 2-butynamide 26 was
significantly deactivated toward thiol addition, exhibiting a t1/2
of 25.5 h.
Scheme 3. t1/2 Values for the Reaction with GSH in
Parentheses
a
Interestingly, MS-guided reaction monitoring of 24 in the
presence of GSH showed an equilibrium of 24 and the adduct
24-GSH was achieved at ∼5 min. This suggested the possibility
of reversible reaction, which is of particular current interest for
the generation of reversible covalent inhibitors.19 To
distinguish a reversible reaction from an ionization-induced
1
artifact, we monitored the reaction of 24 using H NMR
(Figure 2). At 25 and at 37 °C, in 20 mM phosphate buffer
prepared with 10:90 D2O/H2O, only the forward Michael
addition was observed with t1/2 values of 2.1 and 0.32 h,
respectively, to afford exclusively the (Z)-β-mercaptoacrylamide
adduct.20 The difference in t1/2 at 37 °C, measured by NMR
versus MS, was likely a result of using 20 mM phosphate in the
NMR experiment (for optimal line shape and spectral quality),
versus 70 mM phosphate in the MS experiment. Despite this
minor difference, the NMR data strongly suggest that the
addition of GSH to 24 is irreversible under approximate
physiological conditions. In addition, that only the kinetic (Z)-
a
Reactions were monitored by MS at 0.1−1 mM electrophile, 10 mM
glutathione, 70 mM phosphate buffer (pH = 7.4), and 30% CH3CN at
37 °C. Fluoride 21 was synthesized and isolated, but was not stable
on standing.
b
in a spirocyclic framework that has recently seen widespread
use in drug discovery.17 Interestingly, 14 was found to react at a
faster rate (t1/2 = 1.55 h) than parent azetidine 1 (t1/2 = 2.78 h),
possibly due to two inductively withdrawing C−O bonds
connecting the ether of 14 to its azetidine, compared to only
1
Figure 2. H NMR monitoring of GSH reaction with propynamide 24. The addition of GSH irreversibly generates the (Z)-β-mercaptoacrylamide
adduct.20,21
C
DOI: 10.1021/acs.orglett.7b00788
Org. Lett. XXXX, XXX, XXX−XXX

Organic Letters
Letter
product is detected provides strong support for an irreversible
process, since a reversible Michael addition would be expected
to generate the thermodynamic (E)-mercaptoacrylamide.21 To
evaluate potential reversibility with other nucleophiles, 24 was
also exposed to β-mercaptoethanol (BME) and cysteamine.
Both BME and cysteamine reacted rapidly with 24 (BME: t1/2
= 0.27 h, 25 °C; cysteamine: t1/2 = <1 min, 25 °C) and with no
evidence of reversibility. Efforts to compare kinetic data
obtained by 1H NMR reaction-monitoring to the more routine
MS-guided reaction monitoring is of current interest, and work
is ongoing.
Synthetic access to the diverse azetidines presented in this
work was rapid and straightforward due to the wide variety of
azetidine building blocks available commercially. As a
penultimate or antepenultimate step, N-Boc-azetidines were
deprotected with HCl and could be directly acylated, after a
solvent switch, with commercial acyl chlorides or by amidation
with corresponding acrylic or alkynoic acids using T3P.22
Downstream chemistries such as alkyne chlorination23 and
Lindar reduction24 offered a range of additional electrophiles.16
These steps provide azetidines appended with electrophiles
displaying a spectrum of reactivity, as exemplified by the
reactive azetidine 1 (t1/2 = 2.78 h), the less reactive 10 and 23
(t1/2 = 6.50, 10.7 h), and the relatively inert 26 (t1/2 = 25.5 h).
The underrepresentation of N-acryloyl azetidines in probe and
drug discovery could possibly be attributed to the unexpected
(but explicable) reactivity trends presented in this study. These
reactivity data may now be used to interpret kinact/Ki ratios,
extrahepatic clearance, and even bench stability profiles for N-
acryloyl azetidines synthesized against a protein target. While
chemical reactivity is certainly one factor to consider in the
design of covalent inhibitors, it is important to recognize that
the medicinal and pharmaceutical properties of any single
molecule will require rigorous empirical evaluation, since
whole-molecule behavior exhibits complexities that remain
unpredictable.
Research Experience (SURE) program that supported this
work.
REFERENCES
■
(1) Yoda, H.; Takahashi, M.; Sengoku, T. Azetidine and Its
Derivatives. In Heterocycles in Natural Products; Majumdar, K. C.,
Chattopadhyay, S. K., Eds.; Wiley: New York, 2011; Chapter 2.
(2) Kato, N.; et al. Nature 2016, 538, 344−349.
(3) Ghorai, M. K.; Das, K.; Shukla, D. J. Org. Chem. 2007, 72, 5859−
5862.
(4) (a) Ericksson, B. I.; Carlsson, S.; Halvarsson, M.; Risberg, B.;
Mattsson, C. Thromb. Haemostasis 1997, 78, 1404−1407. (b)
AstraZeneca, PCT Int. Appl., WO00/41716, 2000. (c) Wellington,
K.; Scott, L. J. Drugs 2003, 63, 2613−2621.
(5) Patricelli, M. P.; Janes, M. R.; Li, L. S.; Hansen, R.; Peters, U.;
Kessler, L. V.; Chen, Y.; Kucharski, J. F.; Feng, J.; Ely, T.; Chen, J. H.;
Firdaus, S. J.; Babbar, A.; Ren, P.; Liu, Y. Cancer Discovery 2016, 6,
316−329.
(6) (a) Bauer, R. A. Drug Discovery Today 2015, 20, 1061−1073.
(b) Singh, J.; Petter, R. C.; Baillie, T. A.; Whitty, A. Nat. Rev. Drug
Discovery 2011, 10, 307−317.
(7) Liu, C.; Achtenhagen, M.; Szostak, M. Org. Lett. 2016, 18, 2375−
2378.
(8) Otani, Y.; et al. J. Am. Chem. Soc. 2003, 125, 15191−15199.
(9) Woodward, R. B. Philos. Trans. R. Soc., B 1980, 289, 239−250.
(10) Suzuki, T.; Kusakabe, J.-I; Kitazawa, K.; Nakagawa, T.;
Kawauchi, S.; Ishizone, T. Macromolecules 2010, 43, 107−116.
(11) Flanagan.; et al. J. Med. Chem. 2014, 57, 10072−10079.
(12) Cee, V. J.; et al. J. Med. Chem. 2015, 58, 9171−9178.
(13) Nonoo, R. H.; Armstrong, A.; Mann, D. J. ChemMedChem 2012,
7, 2082−2086.
(14) Jackson, P. A.; Widen, J. C.; Harki, D. A.; Brummond, K. M. J.
Med. Chem. 2017, 60, 839−885.
(15) A single example of an N-acryloyl azetidine reacting with N-α-
acetyl-lysine was recently reported at pH 10.2, though analysis of
reactivity trends at physiologic pH was not possible due to protonation
of the amine nucleophile. See: Dahal, U. P.; Gilbert, A. M.; Obach, R.
S.; Flanagan, M. E.; Chen, J. M.; Garcia-Irizarry, C.; Starr, J. T.; Schuff,
B.; Uccello, D. P.; Young, J. A. MedChemComm 2016, 7, 864.
(16) See Supporting Information.
(17) Burkhard, J. A.; Wagner, B.; Fischer, H.; Schuler, F.; Muller, K.;
̈
ASSOCIATED CONTENT
■
Carreira, E. M. Angew. Chem., Int. Ed. 2010, 49, 3524−3527.
(18) Alternatively, conformational rigidity imposed by the spirocyclic
framework may also affect the reactivity of 14.
S
* Supporting Information
The Supporting Information is available free of charge on the
ACS Publications website at DOI: 10.1021/acs.or-
glett.7b00788.
(19) Krishnan, S.; Miller, R. M.; Tian, B.; Mullins, R. D.; Jacobson,
M. P.; Taunton, J. J. Am. Chem. Soc. 2014, 136, 12624−12630.
(20) The resonances at 7.46 ppm belonging to the phenyl protons
were used as an internal reference for quantitating the integrals of Ha
and Hb. Olefin geometry was assigned based on the coupling constant,
J, between Ha and Hb (10.2 Hz) and through comparison to previously
characterized (E) and (Z) mercaptoacrylamides, in: Rossi, R.; Bellina,
F.; Mannina, L. Tetrahedron 1997, 53, 1025−1044.
Experimental procedures and characterization data
(PDF)
AUTHOR INFORMATION
■
(21) We favor the mechanistic proposal put forth by Ma et al. to
explain why the (Z) isomer is generated over the (E) isomer. See: Ma,
S.; Lu, X.; Li, Z. J. Org. Chem. 1992, 57, 709−713.
Corresponding Author
*E-mail: rbauer@lilly.com.
ORCID
(22) Wissmann, H.; Kleiner, H. Angew. Chem. 1980, 92, 129−130.
(23) Oakdale, J. S.; Sit, R. K.; Fokin, V. V. Chem. - Eur. J. 2014, 20,
11101−11110.
Renato A. Bauer: 0000-0001-5352-6389
Notes
(24) Lindlar, H. Helv. Chim. Acta 1952, 35, 446−450.
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank the following individuals at Eli Lilly: Brian Watson
and Jacqueline Wurst for helpful discussions and Robert Boyer,
Christopher Reutter, Jack Calvert, and Paul McDermott for
helpful analytical expertise. Steven Green is gratefully acknowl-
edged for organizing the Eli Lilly Summer Undergraduate
D
DOI: 10.1021/acs.orglett.7b00788
Org. Lett. XXXX, XXX, XXX−XXX