Intramolecular Ring-Opening of Oxetanes: Access to Functionalised Hydroxymethyl 2,3-Dihydroimidazo[1,2- c ]quinazolines

Article abstract of DOI:10.1055/s-0039-1691578

An intramolecular oxetane ring-opening was developed, affording novel 2-(hydroxymethyl)-2,3-dihydroimidazo[1,2- c ]quinazolines from N -(3-methyloxetan-3-yl)quinazolin-4-amines under mild conditions. The resulting medicinally relevant tricyclic scaffolds were synthesised in good yields with diverse substituents. Moreover, reaction optimisation led to the development of a one-pot procedure.

Full text of DOI:10.1055/s-0039-1691578

SYNLETT
0
9
3
6
-
5
2
1
4
1
4
3
7
-
2
0
9
6
© Georg Thieme Verlag Stuttgart · New York
2020, 52, 502–506
letter
502
en
Synlett
Letter
O. R. Steward et al.
Intramolecular Ring-Opening of Oxetanes: Access to Function-
alised Hydroxymethyl 2,3-Dihydroimidazo[1,2-c]quinazolines
Sharan K. Bagala
Michael S. Bodnarchuka
Thomas A. Kinga
Darren McKerrechera
Xuehong Luob
Peng Wangb
Oliver R. Steward*a
a Chemistry, Oncology R&D, AstraZeneca, Cambridge, UK
oliver.steward@astrazeneca.com
b Pharmaron Beijing Co., Ltd., 6 Taihe Road BDA, Beijing,
100176, P. R. of China
Received: 22.10.2019
Accepted after revision: 23.12.2019
Published online: 29.01.2020
(aziridin-1-yl)quinazoline and sodium iodide at reflux has
also been discussed, and this involves a ring-opening intra-
molecular cyclisation (Scheme 1c).7
DOI: 10.1055/s-0039-1691578; Art ID: st-2019-v0577-l
Abstract An intramolecular oxetane ring-opening was developed, af-
fording novel 2-(hydroxymethyl)-2,3-dihydroimidazo[1,2-c]quinazo-
lines from N-(3-methyloxetan-3-yl)quinazolin-4-amines under mild con-
ditions. The resulting medicinally relevant tricyclic scaffolds were
synthesised in good yields with diverse substituents. Moreover, reaction
optimisation led to the development of a one-pot procedure.
Previous work
a) A. F. M. Vlasenko, 1974
H
N
OH
2
N
i) SOCl2 or POCl3, reflux, 2 h
ii) NaOH (10%), reflux, 2 h
10
7
3
9
8
N
N
5
N
N
DHQ
Key words oxetane ring-opening, intramolecular cyclization, imidazo-
quinazolines, one-pot synthesis
b) E. E. S. Korshin, 2012
N
CH(OMe)3
2,3-Dihydroimidazo[1,2-c]quinazolines (DHQs) contain
two six-membered aryl groups and a fused dihydroimidaz-
ole group. The scaffold has an embedded 4-aminoquinazo-
line moiety; these are known kinase inhibitors in drugs. For
example, gefitinib is a known epidermal growth factor re-
ceptor inhibitor in which the N1-quinazoline nitrogen par-
ticipates in a key interaction with the protein and the 4-
amino group directs a substituent into the hydrophobic
pocket.1–3 With the desire to explore new chemical space,
and based on our work exploring new quinazoline-related
motifs, we report a novel reaction to access functionalised
DHQs based on the ring-opening of N-(3-methyloxetan-3-
yl)quinazolin-4-amine through an intramolecular 5-exo-tet
cyclisation under mild conditions.4
The synthesis of DHQs through cyclisation of 2-
[(quinazolin-4-yl)amino]ethan-1-ol in thionyl chloride or
phosphorus chloride has been previously described
(Scheme 1a).5 More recently, a cyclisation of 2-(4,5-dihy-
dro-1H-imidazol-2-yl)aniline and trimethyl orthoformate
at high temperature has also been reported (Scheme 1b).6
Interestingly, the formation of the tricyclic system from 4-
N
115–120 °C, 12 h
N
N
H
NH
N
c) M. Grifantini, 1974
R1 R2
R1
N
N
acetone
R2
NaI, reflux, 3 h
N
N
R1 = H, Me, Ph
R2 = H, Me, Ph
(6 examples)
N
N
d) This work
O
OH
H
N
N
polar solvent
reflux, 2–16 h
R1
R2
R1
R2
N
N
N
N
3a–f
4a–f
Scheme 1 Previously reported syntheses of 2,3-dihydroimidazo[1,2-
c]quinazolines
© 2020. Thieme. All rights reserved. Synlett 2020, 52, 502–506
Georg Thieme Verlag KG, Rüdigerstraße 14, 70469 Stuttgart, Germany

503
O. R. Steward et al.
Letter
Synlett
The ring-opening of strained heterocycles such as oxe-
tanes has been widely explored in the literature.8 Further-
more the ring-strain of an oxetane has been exploited as a
versatile synthetic handle. For example, in the total synthe-
sis of gelsemine, an oxetane underwent nucleophilic ring-
opening with a urethane nitrogen under Lewis acid activa-
tion.9
The reaction was sensitive to the electronic nature of
the quinazoline ring, and electron-withdrawing groups at
R1 gave faster conversion, whereas substrates with elec-
tron-donating groups were slower to react. Full conversion
of 4-chloroquinazolines with electron-donating groups was
achieved by using higher temperatures and longer reaction
time. Substrates with 6-methoxy or 6-iodo functional
groups (1c and 1d, respectively) required longer heating to
achieve full consumption of the starting material. To our
surprise, however, prolonged heating did not deliver the ex-
pected aminoquinazolines 3, but a cyclisation event oc-
curred under these conditions to generate the correspond-
ing imidazoquinazolines 4c and 4d.
We next explored the oxetane ring-opening step of in-
termediate 3f (Table 2). Complete conversion was observed
on heating at reflux in methanol for 16 hours (Table 2, entry
1). We surmised that use of a higher boiling solvent might
increase the rate of conversion. However, when 1,4-dioxane
was used at 100 °C (entry 2), no product was obtained.
When methanol was added to the same reaction mixture,
product 4f was obtained. These results indicated that
changing to a polar protic solvent with a high boiling point
might be beneficial. To this end, we used propa-1-ol. Grati-
fyingly, full conversion was achieved within four hours (en-
try 3). By changing the method of heating to microwave ir-
radiation, the reaction time in polar protic solvents could be
reduced further (entries 4 and 5).
Functionalising the 2-position of DHQs is attractive
from the point of view of drug discovery, as this process can
be used to explore new vectors of three-dimensional space
in both saturated and unsaturated scaffolds. Moreover, to
the best of our knowledge, we are unaware of any other
methods for installing a functional handle for subsequent
manipulation at the 2-position in analogous systems.
Our strategy was to prepare the 2-hydroxymethyl DHQs
4a–f from the corresponding N-(3-methyloxetan-3-
yl)quinazolin-4-amines 3a–f with diverse substitu-
tion/electronics to probe the oxetane ring-opening reac-
tion. We envisaged that these novel compounds might be
manipulated further through the primary alcohol group.
We began our study with the preparation of N-(3-methyl-
oxetan-3-yl)quinazolin-4-amines 3a–f from the corre-
sponding 4-chloroquinazolines 1a–f. 4-Chloroquinazolines
1a–f were purchased from commercial sources or were pre-
pared from the corresponding 4-hydroxyquinazoline by re-
fluxing in thionyl chloride and N,N-dimethylformamide for
4–16 hours. The isolated 4-chloroquinazolines were subse-
quently treated with 3-methyloxetan-3-amine (2) and po-
tassium carbonate in acetonitrile. Table 1 shows the reac-
tion conditions used to achieve full conversion of 4-chloro-
quinazolines 1a–f into the N-(3-methyloxetan-3-
yl)quinazolin-4-amines 3a–f.
This led to an exploration of the scope of this transfor-
mation from intermediates 3a–f under our optimised con-
ditions (Table 3). We observed the general trend that sub-
strates with electron-withdrawing groups in the R1 position
Table 1 Formation of N-(3-Methyloxetan-3-yl)quinazolin-4-amines and (2-Methyl-2,3-dihydroimidazo[1,2-c]quinazolin-2-yl)methanols
O
H
Cl
N
N
K2CO3 (2 equiv)
MeCN
R1
R2
R1
R2
O
N
N
+
NH2
N
1a–f
2
3a–f
Entry
R1
CO2Et
R2
Temp (°C)
Time (d)
Product
Yield (%)
1a
1b
1c
1d
1e
1f
H
H
H
H
H
Br
50
50
90
90
90
90
1
1
3
2
2
2
3a
3b
4cc
4dd
3e
3f
92a
59b
31
NO2
OMe
I
32
Me
Cl
76a
78a
a No chromatography was required.
b Normal-phase chromatography was used.
c Full conversion into 4c was observed.
d Full conversion into 4d was observed.
© 2020. Thieme. All rights reserved. Synlett 2020, 52, 502–506

504
O. R. Steward et al.
Letter
Synlett
Table 2 Formation of (8-Bromo-9-chloro-2-methyl-2,3-dihydroimid-
azo[1,2-c]quinazolin-2-yl)methanol (4f)
porated on the newly formed hydroxy group before oxida-
tion using KMnO4.7 The TBDMS group was then removed by
using TBAF to afford the desired imidoquinazole 10.
O
OH
H
N
N
O
H
Cl
Br
Cl
Br
Cl
N
TEA
N
N
N
O
+
Cl
Br
Cl
Br
N
N
N
N
MeCN, 40 °C, 3.5 h
N
H2
3f
4f
N
1a
5
6
(94% yield)
Entry
Solvent
Temp (°C)
Time (h)
16
OH
1
2
3
4
5
MeOH
70
N
TBDMSCl
imidazole
a
MeOH
1,4-Dioxane
PrOH
100
100
100b
100b
–
Cl
Br
N
100 °C, 9 h
(microwave)
4
DMF, 25 °C, 2.5 h
N
MeOH
1
7
(70% yield)
PrOH
0.75
OTBDMS
OH
a No conversion observed.
b Microwave heating.
i) KMnO4, silica gel
MeCN, 25 °C, 18 h
52% yield (9)
N
N
N
Cl
Br
Cl
N
ii) TBAF,
THF, 25 °C, 15 min
N
Br
N
afforded higher yields at lower temperatures, as compared
with those with electron-donating groups in this position,
when methanol was used as the solvent.
8
10
(98% yield)
(28% yield)
Scheme 2 Synthetic route to (8-bromo-9-chloro-2,3-dihydroimid-
azo[1,2-c]quinazolin-2-yl)methanol (10)
Table 3 Formation of 2-Hydroxymethyl DHQs
O
OH
On the basis of Hammett reactivity, we expected elec-
tron-donating groups to enhance the nucleophilicity of the
quinazoline nitrogen.10 Instead, we observed the reverse;
we therefore performed quantum-mechanical (QM) calcu-
lations on the transformation of 3b into 4b and of 3e into 4e
to help to understand the reaction mechanism and the vari-
ous yields observed. The QM calculations indicated that
both reactions afford a similar stabilised intermediate,
shown in Figure 1, through attack by the quinazoline N3
atom on the oxetane methylene group in an SN2-like man-
ner. In the stabilised intermediate, the molecule is predict-
ed to exist as a zwitterion with the quinazoline N3 atom
carrying a positive charge and with a negative charge on
the alkoxide stabilised by the amino group. Such an obser-
vation is consistent with the need for a polar protic solvent
in the reactions.
H
N
N
R1
R2
R1
R2
MeOH
N
N
N
N
4a–f
3a–f
Product R1
R2
Temp (°C)
Time (h)
Yield (%)
4a
4b
4c
4d
4e
4f
CO2Et
NO2
OMe
I
H
70
100a
–
16
2
62
H
H
H
H
Br
93
–
31b
32b
20
–
–
Me
Cl
100a
70
2
16
35
a Microwave heating.
b Final yield after conditions from Table 1.
We next decided to use oxetan-3-amine (5) to synthe-
sise a des-methyl DHQ tricycle with a view to making a ful-
ly aromatic system through oxidation of the CH(CH2OH)–
CH2 bond (Scheme 2). The initial SNAr reaction using ox-
etan-3-amine (5) proceeded in excellent yield. Switching to
an organic base such as triethylamine improved the workup
procedure, and the subsequent cyclisation was completed
by using microwave conditions to afford (8-bromo-9-
chloro-2,3-dihydroimidazo[1,2-c]quinazolin-2-yl)metha-
nol (7) in high yield. A TBDMS protecting group was incor-
Figure 1 QM-predicted geometries of 3b (cyan), 4b (green), and the
stabilised intermediate state (yellow)
The QM-minimised single-point energy (SPE) of the sta-
bilised intermediates was used to define the activation en-
ergy (Ea) of the reaction for both 4b and 4e, whereas the re-
© 2020. Thieme. All rights reserved. Synlett 2020, 52, 502–506

505
O. R. Steward et al.
Letter
Synlett
action energy (Er) was defined by comparing the QM SPE of
the product and the initial reactant. These energies for 4b
and 4e are shown in Table 4, and they indicate that the
presence of a 6-nitro functional group dramatically stabilis-
es the intermediate relative to that of a 6-methyl functional
group, as well as affording a product which is slightly lower
in energy. Such values are fully consistent with the ob-
served yields in Table 3. This suggests that upon heating,
the cyclisation occurs more quickly for substrates with elec-
tron-withdrawing groups than for those with electron-do-
nating groups, due to the reduced activation energy and
improved stabilisation of the final product.
formed a one-pot process reaction at 50 °C and three one-
pot process reactions at 150 °C under identical reaction
conditions and monitored the time required for completion
(Table 5). We observed only the SNAr product 3b by LCMS
after heating at 50 °C for one hour, whereas after heating at
150 °C for one hour, 4b was observed, demonstrating that
the rate-determining step is dependent on the reaction
temperature after insertion of the amino-oxetane. For elec-
tron-withdrawing groups, such as 6-nitro, conversion of 1b
into 4b as the major product was observed with a shorter
reaction time (1 h) compared with electron-donating
groups, such as 6-methoxy (4 h for full consumption of 1c).
This is consistent with our observations shown in Table 1
and our QM-based reaction analysis (Table 4).
Table 4 Activation Energies (Ea) and Reaction Energies (Er) for 3b to 4b
and 3e to 4e
Table 5 Formation of 2-Hydroxymethyl DHQs by Using a One-Pot Processa
Conversion
R1
Ea (kcal/mol)
Er (kcal/mol)
Yield (%)
O
O
OH
3b into 4b
3e into 4e
NO2
Me
27.5
43.0
–19.2
–18.4
93
20
H
Cl
N
N
N
N
2
N
R1
R1
R1
H2
DIPEA, MeCN
N
N
N
N
On the basis of the results shown in Table 4, a plausible
mechanism was proposed (Scheme 3). Given that a polar
protic solvent produces rate enhancement, it is conceivable
that the solvent plays a role in the mechanism through sta-
bilisation of the intermediate species or by acting as a hy-
drogen-bond donor. We observed that higher temperatures
are required when electron-donating groups are present on
the quinazoline, as these substrates have higher activation
energies that must be overcome to enable transformation
from intermediate shown in (b) to that shown in (c) in
Scheme 3. This explains why 1c (Table 1) was converted di-
rectly into the ring-opened product 4c without isolation of
intermediate 3c.
3b,c
4b,c
1b,c
Entry Conversion R1
Time (h) Temp (°C) Conversion (1/3/4/U)b
1
2
3
1b into 3b
1b into 4b
1c into 4c
NO2
NO2
OMe
1
1
4
50
150
150
0:79:0:21
0:0:71:29
0:0:80:20 (71% yield)
a Reaction conditions: 2 (1.2 equiv), DIPEA (2 equiv), microwave irradia-
tion11
b U = unknown biproduct.
In conclusion, a novel route to hydroxymethyl 2,3-dihy-
droimidazo[1,2-c]quinazolines through the nucleophilic
ring-opening of an oxetane in an intramolecular 5-exo-tet
cyclisation has been developed. A range of novel 2,3-dihy-
droimidazo[1,2-c]quinazolines were produced from either
3-methyloxetan-3-amine or oxetan-3-amine. The reaction
proceeded by heating in polar solvents such as methanol,
propan-1-ol, or acetonitrile with thermal or microwave
heating, and was used to produce a range of 6-function-
alised compounds with a range of electronic properties. Fi-
nally, the complementarity of the SNAr and cyclisation steps
allowed a one-pot procedure to be developed, resulting in
rapid construction of molecular complexity from simple
starting materials. It is predicted that the intramolecular
cyclisation results in the formation of an intermediate sta-
bilised by zwitterionic interactions. QM-minimised SPEs
were also calculated to reveal activation and reaction ener-
gies that supported the hypothesis that the cyclisation of
electron-deficient quinazolines occurred faster than that of
electron-rich quinazolines, as observed experimentally.
Given the need to develop novel binding motifs and to
explore new chemical space, we believe this methodology
should be of interest to both academic and industrial re-
O–
a)
b)
H
H
N+
O
N
N
N
N
N
O–
OH
c)
H
N
N
N+
N
N
N
Scheme 3 Proposed mechanism based on QM calculations
For entries such as 1b (Table 1), a lower temperature
was used for the initial SNAr reaction and consequently no
ring-opening was observed. To test this hypothesis, we per-
© 2020. Thieme. All rights reserved. Synlett 2020, 52, 502–506

506
O. R. Steward et al.
Letter
Synlett
searchers. Future work will include investigating the ring-
opening reaction to synthesise six-membered rings and
other heterocyclic compounds.
(6) Korshin, E. E.; Sabirova, L. A.; Levin, Y. A. Synthesis 2012, 44,
3512.
(7) Claudi, F.; Franchetti, P.; Grifantini, M.; Martelli, S. J. Org. Chem.
1974, 39, 3508.
(8) Bull, J. A.; Croft, R. A.; Davis, O. A.; Doran, R.; Morgan, K. F. Chem.
Rev. 2016, 116, 12150.
(9) Ng, F. W.; Lin, H.; Tan, Q.; Danishefsky, S. J. Tetrahedron Lett.
2002, 43, 545.
Funding Information
Funding sources from AstraZeneca.Astar
Z
e
n
e
c
a
()
(10) Hammet, L. P. J. Am. Chem. Soc. 1937, 59, 96.
(11) (2-Methyl-2,3-dihydroimidazo[1,2-c]quinazolin-2-yl)metha-
nols 4a–f; General Procedure
Acknowledgment
The appropriate 4-chloroquinazoline 1 (100 mg), 3-methylox-
etan-3-amine (2; 1.2 equiv), and DIPEA (2 equiv) were dissolved
in MeCN (3 mL) and the solution was sealed in a microwave
tube. The tube was purged with N2 for 1 min and then heated to
150 °C for 4 h in a microwave reactor. The mixture was cooled
to 25 °C then evaporated to dryness and purified by preparative
HPLC (Waters XSelect CSH C18 ODB column, 5 silica, 30 mm
diameter, 100 mm long), eluting with decreasingly polar mix-
tures of 1% aq NH3 and MeCN. Fractions containing the desired
compound were evaporated to dryness to afford the product 4.
(2-Methyl-9-nitro-2,3-dihydroimidazo[1,2-c]quinazolin-2-
yl)methanol (4b)
We would like to thank Scott Martin for acquiring high-resolution
mass spectroscopy data.
Supporting Information
Supporting information for this article is available online at
https://doi.org/10.1055/s-0039-1691578.
S
u
p
p
orti
n
gInformati
o
n
S
u
p
p
orit
n
gInformati
o
n
References and Notes
Yellow solid; yield: 288 mg (93%). 1H NMR (400 MHz, DMSO-
d6):  = 1.24 (s, 3 H, Me), 3.36 (dd, J = 6.4, 10.9 Hz, 1 H, CH2O),
3.44 (dd, J = 5.9, 10.9 Hz, 1 H, CH2O), 3.77 and 4.15 (d, J = 10.9
Hz, 1 H, and d, J = 10.9 Hz, 1 H, CH2), 5.00 (t, J = 5.9 Hz, 1 H, OH),
7.53 (d, J = 8.9 Hz, 1 H, Ar-H), 8.10 (s, 1 H, Ar-H), 8.32 (dd, J = 2.7,
8.9 Hz, 1 H, Ar-H), 8.56 (d, J = 2.7 Hz, 1 H, Ar-H). 13C NMR (101
MHz, DMSO-d6, 30 °C):  = 24.27 (CH3), 53.81 (CH2), 67.66
(CH2O), 71.42 (C), 118.57 (C), 120.54 (CH), 127.07 (CH), 127.93
(CH), 144.18 (C), 147.86 (CH), 150.12 (C), 151.49 (C). HRMS
(ESI): m/z [M + H]+ calcd for C12H13N4O3: 261.0909; found:
261.0988, 215.1053 [M – NO2]+.
(1) Kettle, J. G.; Wilson, D. M. Drug Discovery Today 2016, 21, 1596.
(2) Wu, P.; Nielsen, T. E.; Clausen, M. H. Trends Pharmacol. Sci. 2015,
36, 422.
(3) Tiwary, B.; Pradhan, K.; Nanda, A. K.; Chakraborty, R. J. Chem.
Biol. Ther. 2015, 1, 104; DOI: 10.4172/2572-0406.1000104.
(4) Baldwin, J. E. J. Chem. Soc., Chem. Commun. 1976, 734.
(5) Vlasenko, A. F.; Mandrichenko, B. E.; Rogul’chenko, G. K.; Sinyak,
R. S.; Mazur, A. I.; Kochergin, P. M. Khim. Geterotsikl. Soedin.
1976, 6, 834.
© 2020. Thieme. All rights reserved. Synlett 2020, 52, 502–506