Synthesis and biological evaluation of quinazoline derivatives – A SAR study of novel inhibitors of ABCG2

Article abstract of DOI:10.1016/j.ejmech.2018.10.026

Multidrug resistance (MDR) is a major obstacle for effective chemotherapeutic treatment of cancer frequently leading to failure of the therapy. MDR is often associated with the overexpression of ABC transport proteins like ABCB1 or ABCG2 which efflux harmful substances out of cells at the cost of ATP hydrolysis. One way to overcome MDR is to apply potent inhibitors of ABC transporters to restore the sensitivity of the cells toward cytostatic agents. This study focusses on the synthesis and evaluation of novel 2,4-disubstituted quinazoline derivatives regarding the structure-activity-relationship (SAR), their ability to reverse MDR and their mode of interaction with ABCG2. Hence, the inhibitory potency and selectivity toward ABCG2 was determined. Moreover, the intrinsic cytotoxicity and the reversal of MDR were investigated. Interaction type studies with the substrate Hoechst 33342 and conformational analyses of ABCG2 with 5D3 monoclonal antibody were performed for a better understanding of the underlying mechanisms. In our study we could further enhance the inhibitory effect against ABCG2 (compound 31, IC₅₀: 55 nM) and identify the structural features that are crucial for inhibitory potency, the impact on transport activity and binding to the protein.

Full text of DOI:10.1016/j.ejmech.2018.10.026

Accepted Manuscript
Synthesis and biological evaluation of quinazoline derivatives – A SAR study of novel
inhibitors of ABCG2
Michael K. Krapf, Jennifer Gallus, Anna Spindler, Michael Wiese
PII:
S0223-5234(18)30888-2
10.1016/j.ejmech.2018.10.026
EJMECH 10812
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 28 August 2018
Revised Date: 8 October 2018
Accepted Date: 10 October 2018
Please cite this article as: M.K. Krapf, J. Gallus, A. Spindler, M. Wiese, Synthesis and biological
evaluation of quinazoline derivatives – A SAR study of novel inhibitors of ABCG2, European Journal of
Medicinal Chemistry (2018), doi: https://doi.org/10.1016/j.ejmech.2018.10.026.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT
Graphical Abstract

ACCEPTED MANUSCRIPT
Synthesis and Biological Evaluation of Quinazoline
Derivatives – A SAR Study of Novel Inhibitors of ABCG2
Michael K. Krapf, Jennifer Gallus, Anna Spindler, Michael Wiese*
Pharmaceutical Institute
University of Bonn
An der Immenburg 4
5
3121 Bonn, Germany
Keywords
ABC transporter; inhibitor; ABCG2; quinazolines; SAR
*
E-mail: mwiese@uni-bonn.de
1

ACCEPTED MANUSCRIPT
Abstract
Multidrug resistance (MDR) is a major obstacle for effective chemotherapeutic treatment of
cancer frequently leading to failure of the therapy. MDR is often associated with the
overexpression of ABC transport proteins like ABCB1 or ABCG2 which efflux harmful
substances out of cells at the cost of ATP hydrolysis. One way to overcome MDR is to apply
potent inhibitors of ABC transporters to restore the sensitivity of the cells toward cytostatic
agents. This study focusses on the synthesis and evaluation of novel 2,4-disubstituted
quinazoline derivatives regarding the structure-activity-relationship (SAR), their ability to
reverse MDR and their mode of interaction with ABCG2. Hence, the inhibitory potency and
selectivity toward ABCG2 was determined. Moreover, the intrinsic cytotoxicity and the
reversal of MDR were investigated. Interaction type studies with the substrate Hoechst 33342
and conformational analyses of ABCG2 with 5D3 monoclonal antibody were performed for a
better understanding of the underlying mechanisms. In our study we could further enhance the
inhibitory effect against ABCG2 (compound 31, IC : 55 nM) and identify the structural
5
0
features that are crucial for inhibitory potency, the impact on transport activity and binding to
the protein.
1
. Introduction
ATP-Binding cassette (ABC) transport proteins are found among eukaryotic and prokaryotic
organisms [1]. In eukaryota they are expressed in different tissues where they provide
membrane transport of nutrients and are responsible for an active efflux of different molecules
upon ATP hydrolysis, protecting the cells from potentially harmful compounds [2,3]. Human
ABC transport proteins form a superfamily of 48 members, classified into seven subfamilies
2

ACCEPTED MANUSCRIPT
ABC A to ABC G [4,5,6]. The Breast Cancer Resistance Protein (BCRP, ABCG2) for
instance is expressed amongst others in tissues like the gut, bile, canaliculi, placenta, blood-
testis, and blood-brain barriers excreting and limiting absorption of potentially harmful
substrate molecules [7,8]. ABCG2 is one of three major members of the ABC transport
protein family besides P-glycoprotein (P-gp, ABCB1) and multidrug resistance associated
protein 1 (MRP1, ABCC1), which are often associated with the occurrence of multidrug
resistance (MDR) in cancer [4,9,10,11]. Breast Cancer Resistance Protein was discovered in
1
998 by Doyle et al. when they selected a human breast cancer cell line for doxorubicin
resistance in the presence of the ABCB1 inhibitor verapamil [12,13]. More cytostatic drugs
like the anthracene derivative mitoxantrone (MX) or the active metabolite of irinotecan,
namely SN-38, were identified as substrates of ABCG2 [ 14 , 15 ]. Therefore, a
chemotherapeutic therapy might lack effectiveness due to overexpression of ABCG2, leading
to a low intracellular concentration of the cytostatic. High expression rates of ABCG2 were
for instance found in solid tumors and hematopoietic tumors [16]. A possible way to
resensitize such cancer cells could be the co-administration of potent inhibitors of ABCG2.
The first potent inhibitor was discovered and isolated from Aspergillus fumigatus, and named
fumitremorgin C (FTC). Despite its high potency, FTC never found clinical application owing
to neurotoxic side effects [17]. More promising was the second generation inhibitor Ko143,
exhibiting a high potency together with selectivity in the submicromolar range toward
ABCG2 [18,19]. The tyrosine kinase inhibitor gefitinib used in clinical application for the
treatment of non-small cell lung carcinoma (NSCLC). It was found that co-administration
with topoisomerase I inhibitors was an effective approach to overcome the resistance of
ABCG2 overexpressing cells [20]. However, only few potent and selective inhibitors of
ABCG2 have been described yet [21].
Our workgroup decided to modify the structure of gefitinib, to increase its relatively low
inhibitory potency toward ABCG2. First step was to start with the underlying quinazoline
3

ACCEPTED MANUSCRIPT
scaffold introducing a phenyl moiety at position 2 und to introduce various substituents at the
aniline linker at position 4. First results from the Hoechst 33342 accumulation assay with
ABCG2 overexpressing cells were promising, yielding several compounds with higher
potencies than Ko143, one of the most potent inhibitor of ABCG2 known to date
[22,23,24,25,26]. In our most recent study we investigated a modification of the quinazoline
scaffold using a pyrido[2,3-d]pyrimidine structure [27]. We found an enantiotopic relation of
the inhibitory activities between those scaffolds leading to several highly potent derivatives.
Also, the intrinsic cytotoxicities of these new compounds were low.
In the present study, we decided to investigate different modifications of our basic 4-anilino-
2
-phenylquinazoline scaffold. We replaced the aromatic residue at position 2 by five-
membered heteroaromatic moieties as well as hydrogen to determine the impact of the
substitution at this position on the inhibitory activity toward ABCG2. Also, we altered the 4-
amino linker by replacing it by oxygen or sulphur. Furthermore, the linker was replaced by an
amido group. Based on the results from the oxygen and sulphur linkers we methylated the
amino linker to investigate the importance of a H-donor function on inhibitory activity.
With these new insights into the structure-activity relationship (SAR) of this novel class of
inhibitors new tailored inhibitors of ABCG2 can be designed.
2
2
. Results and Discussion
.1. Chemistry. A brief description of the synthetic route for all compounds is presented in
Scheme 1.
4

ACCEPTED MANUSCRIPT
a
:
Reagents and conditions: (i) DMF, I , K CO , 70-90 °C, 4-8 h. (ii) POCl , reflux, 4-12 h.
2
2
3
3
(iii) Substituted aniline, isopropanol, 100 watt microwave irradiation, 110 °C, 15 - 30 min. (iv)
Substituted amine, isopropanol, 100 watt microwave irradiation, 110 °C, 15 - 30 min; phenol,
thiophenol), ethanol (water), reflux, 5-10 (4) h (v) 4-Substituted-2-phenylquinazoline, MeI,
(
5

ACCEPTED MANUSCRIPT
NaH, DMF, 0 °C 1h, RT 2-6 h. (vi) t-BuOK, 150 watt microwave irradiation, 180 °C, 2 min.
(vii) substituted benzoyl chloride, THF, TEA, RT, 12 h.
Scheme 1. General Synthesis Route for Compounds 1-40.
Quinazolinone derivatives containing a five-membered aromatic ring at position 2 were
synthesized via cyclic condensation from anthranilamide together with thiophene-2-
carbaldehyde or 1H-pyrrole-2-carbaldehyde to yield the corresponding precursor 1 or 2.
Chlorination of the carbonyl function at position 4 was achieved by reaction with POCl to
3
yield 3 and 4. The 2-substituted-4-chloroquinazoline derivatives were then reacted with
different substituted anilines to obtain compounds 5-13 via nucleophilic aromatic substitution
at position 4.
In scaffold B the initial amino linker was replaced by oxygen and sulphur for compounds 15-
2
4. The derivatives were obtained by reaction of 2-phenyl-4-chloroquinazoline with different
substituted phenols and thiophenols. Compounds 25-28 were obtained by methylation of the
2
anilino linker at position 4 of the corresponding derivatives based on scaffold B (R : NH).
2
Moreover, the anilino linker of scaffold B (R : NH) was replaced in scaffold C of compounds
2
9-32 by an amido linker. Initially, the 2-phenyl-4-aminoquinazoline was synthesized by a
cyclization of 2-aminobenzonitrile and benzonitrile undergoing an intermolecular
nucleophilic attack by the amino-function followed by an intramolecular nucleophilic attack
of the formed imine leading to ring closure (14). Compounds 29-32 were prepared from 14
and the corresponding substituted benzoylchloride, forming the amido function by
nucleophilic substitution.
Compounds 33-40 were synthesized by nucleophilic aromatic substitution of 4-
chloroquinazoline and different substituted anilines. This structure shows the closest
6

ACCEPTED MANUSCRIPT
resemblance to gefitinib, both lacking substitution at position 2. However, gefitinib possesses
additional substituents at position 6 and 7.
Presented variations of the moieties based on the basic quinazoline scaffold should contribute
to a better understanding of the SAR regarding quinazoline derivatives as inhibitors of
ABCG2.
1
13
The identity of all compounds was confirmed by H and C NMR spectra and the purity by
elemental analysis. All values found were in the range of ± 0.4% of the theoretical values,
unless indicated.
2
2
.2. Biological evaluation.
.2.1. Hoechst 33342 Accumulation Assay. The inhibitory activity toward ABCG2 was
determined using the Hoechst 33342 accumulation assay. This assay was carried out as
previously described (see experimental section). Once the dye is embedded in a lipophilic
environment like a cell membrane or bound to DNA, fluorescence increases strongly [28].
Being a substrate of ABCG2 the dye can be used as an indicator of intracellular accumulation,
which is dependent on the degree of inhibition of the transport protein. Thus, the measured
fluorescence can be correlated with the inhibitory potency of a substance toward ABCG2.
Results from the Hoechst accumulation assay are presented in Table 1.
Table 1. Inhibitory activities of derivatives in the Hoechst 33342 accumulations assay toward
ABCG2 overexpressing MDCK II BCRP cell line. The scaffold with substitution pattern is
illustrated in the heading.
7

ACCEPTED MANUSCRIPT
Hoechst 33342
1
2
a
e
Compd.
R
R
Scaffold IC ± SD [nM]
Inhibition [%]
50
5
6
7
8
9
3-NO
4-NO
3-CN
2
2
S
S
S
S
S
S
A
A
A
A
A
A
A
A
A
300 ± 47
335 ± 42
178 ± 19
202 ± 21
953 ± 176
360 ± 20
156 ± 10
2570 ± 160
652 ± 137
n.a.
92
84
92
3,4-OMe
3-F
102
86
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
0
1
2
3
4*
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
3-NHCOCH
3-CN
3
90
NH
NH
NH
104
103
98
4-OMe
3,4-OMe
f
H
O
B
B
B
B
B
B
B
B
B
B
B
B
B
B
C
C
C
C
D
D
D
14500 ± 2100
4780 ± 1620
-
f
3-NO
4-NO
3-Br
O
-
2
2
g
f
O
n.a.
-
f
O
3860 ± 960
7270 ± 2080
989 ± 146
>>10000
-
f
4-Br
O
-
f
3-CF
H
O
-
3
f
S
-
g
f
3-Br
3-CF
S
n.a.
-
f
S
14800 ± 4200
7400 ± 3650
2025 ± 187
3669 ± 385
10080 ± 1990
3948 ± 347
424 ± 32
-
3
f
3-OMe
S
-
f
3-NO
4-NO
4-CN
3-F
N-Me
N-Me
N-Me
N-Me
-
2
2
f
-
f
-
f
-
H
59
64
70
61
2-NO
3-NO
4-NO
H
994 ± 99
2
2
2
54.5 ± 9.4
93.1 ± 8.4
f
10700 ± 2100
2900 ± 490
2630 ± 840
-
f
3-NO
4-OH
-4-OH
-
2
55
8

ACCEPTED MANUSCRIPT
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4-CN
D
D
D
D
D
B
B
B
B
B
B
B
B
B
B
B
B
B
1270 ± 460
645 ± 68
75
67
60
3-OMe
3-SMe
3-F
1350 ± 320
8120 ± 2190
28900 ± 4600
882 ± 157
130 ± 30
f
-
f
3-NHCOCH
H
-
3
b
c
b
c
c
c
c
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
NH
96
3-NO
4-NO
3-NO
4-OH
3-CN
4-CN
n.d.
84
2
2
2
69.6 ± 8.0
80.0 ± 9.1
204 ± 37
-4-OH
81
87
140 ± 40
n.d.
87
69.9 ± 10
1320 ± 100
1930 ± 110
152 ± 19
c
3-OMe
4-OMe
3,4-OMe
3-SMe
n.d.
n.d.
97
c
b
c
c
c
1190 ± 31
355 ± 53
86
3-F
84
3-NHCOCH
278 ± 33
98
3
d
Ko143
227 ± 14
100
88
d
gefitinib
1730 ± 270
a
:
:
:
:
:
IC values are means of three independent experiments.
50
b
Compound was first synthesized elsewhere [23].
Data taken from our previous studies [24].
Used as reference in the corresponding assay.
Percentage of inhibition with regard to Ko143
Top value fixed to Ko143
c
d
e
f
:
g
:
Poorly soluble
n.a.: Not active.
n.d.: No data available.
*
:
See substitution pattern.
For a better clarity we summarized the SAR of all modifications including their corresponding
compounds with the highest potency in Figure 1.
9

ACCEPTED MANUSCRIPT
Figure 1. SAR of some substitution patterns used in the different quinazoline derivatives.
The IC value of the most potent compounds determined in the Hoechst 33342 accumulation
5
0
assay is shown below the corresponding structure.
In our first series, a thienyl moiety was introduced at position 2 of the quinazoline scaffold to
compare the five-membered heterocycle with a phenyl residue, which has been intensively
1
investigated in earlier studies [22,23,24,26]. As substituents at R , several groups with
different chemical properties were investigated:
1
At first, a nitro group was introduced at R in meta and para position obtaining compound 5
and 6, respectively. The IC values of 300 nM and 335 nM illustrated, that substitution with
5
0
1
thienyl at R yielded less potent compounds than with phenyl (42: IC : 130 nM; 43: IC :
5
0
50
6
9.6 nM). Nevertheless, substitution with meta cyano resulted in a high inhibitory potency
with an IC of 178 nM for compound 7. The 2-phenyl analogue 46 possessed a similar IC
5
0
50
value of 140 nM. Also the 3,4-dimethoxy derivative 8 was highly potent with an IC of 202
5
0
nM. Again, it was slightly less potent than its phenyl analogue 50 (IC : 152 nM). The lowest
5
0
1
potency among the thienyl derivatives was obtained with a 3-fluoro substituent at R (9; IC :
5
0
9
53 nM), although still in the submicromolar range. The corresponding phenyl analogue 52
on the other hand possessed an IC of 355 nM. The 3-acetamido derivative 10 was found to
5
0
1
0

ACCEPTED MANUSCRIPT
be almost equally potent as its phenyl analogue 53, with IC values of 360 nM and 278 nM,
5
0
respectively. Moreover, some derivatives with a pyrrolyl residue at position 2 were
synthesized and tested to find similarities and differences to the substitutions with thienyl or
1
phenyl. Hence, a 3-cyano residue was introduced at R to yield compound 11. The IC value
5
0
of 156 nM is comparable to both, the thienyl and the phenyl analogue and revealed the highest
inhibitory potency among the five-membered heterocycles. The concentration-response curve
of compound 11 with Ko143 is depicted in Figure 2.
100
80
60
40
20
0
-
-8
-7
-6
concentration log [M]
Figure 2. Exemplary concentration-response curve of compound 11 (open circle; IC : 156
5
0
nM) and 31 (open square; IC : 54.5 nM) in a Hoechst 33342 accumulation assay with Ko143
5
0
(
closed diamond, IC : 227 nM) as reference, using the ABCG2 overexpressing MDCK II cell
50
line.
A low inhibitory potency was found for compound 12, with a 4-methoxy substitution. The
resulting IC value of 2570 nM is again higher than for the phenyl analogue 49 (IC : 1930
5
0
50
nM). Likewise, substitution with 3,4-dimethoxy (13; IC : 652 nM) exhibits a higher IC
50
5
0
value than its phenyl analogue 50 (IC : 152 nM). In the next series, the anilino linker at
5
0
position 4 was replaced by an oxygen or sulphur linker. The compounds 15-24 were found to
be poor inhibitors with IC values in the micromolar range. This was somewhat unexpected
50
1
1

ACCEPTED MANUSCRIPT
and pointed to the importance of NH for inhibitory potency. Therefore, we treated some of
2
our derivatives based on scaffold B (R : NH) with methyl iodide to perform a methylation of
the aniline linker at position 4. This also led to a significant decrease in the inhibitory potency
of the compounds substantiating the assertion that the NH-function is highly favourable.
Methylated derivatives were at least 11-fold less potent than their counterparts (see
compounds 25-28 vs. 42, 43, 47 and 52). In terms of inhibitory potency the presence of a H-
donor function seems to be highly preferable. If this is absent the methylamino linker is
preferred over an oxygen linker while the sulphur linker performs worst.
Moreover the aniline linker was replaced by an amido linker. For this purpose the precursor
1
4 with a free amino function at position 4 was investigated first. However, no inhibitory
activity could be measured in the Hoechst 33342 accumulation assay. Though, the
unsubstituted compound 29 with amido linker possessed an IC of 424 nM, a considerably
5
0
two-fold higher potency than its anilino analogue 41 (IC : 882 nM). Regarding the
5
0
percentage of inhibition in comparison to Ko143, all investigated compounds containing an
amido linker achieved a maximal inhibition in the range of 60-70%. Nevertheless, high
inhibitory potencies were achieved for some modifications. As 2-phenylquinazolines with
substitution of meta or para nitro anilines were found to be potent inhibitors of ABCG2, ortho,
1
meta and para nitro functions were investigated on scaffold C at R to yield compounds 30,
3
3
1 and 32. Except compound 30 (IC : 994 nM), the meta and para nitro derivatives 31 and
50
2 possessed extraordinary low IC50 values of 54.5 nM and 93.1 nM, respectively.
Subsequently, the importance of the phenyl moiety at position 2 of the quinazoline scaffold
was investigated. Substitution was carried out at the aniline-linker at position 4 (see scaffold
D). Besides the lack of substitution at position 6 and 7, this class of compounds has the
highest resemblance with the tyrosine kinase inhibitor gefitinib (see Figure 3; IC : 1730 nM).
5
0
1
2

ACCEPTED MANUSCRIPT
Figure 3. Structural formula of the tyrosine kinase inhibitor gefitinib. In the Hoechst 33342
assay an IC of 1730 nM was determined.
5
0
First, the unsubstituted compound 33 was investigated, resulting in an IC50 of 10700 nM.
Inhibitory potency could be enhanced by using a 3-nitro-4-hydroxy residue (34; IC : 2900
5
0
1
nM) or a 4-hydroxy group at R (35; IC : 2630 nM). A moderately high inhibitory activity
5
0
among derivatives with scaffold D was found for compound 36 (IC : 1270 nM), with a 4-
5
0
cyano substituent. Comparison with their phenyl analogues 41, 44, 45 and 47 illustrates, that
the presence of a phenyl or other aromatic moiety is important for the inhibitory potency of a
compound. Good inhibitory activities were found for 3-methoxy and 3-thiomethyl substituted
1
compounds at R of scaffold D. Resulting IC values of 645 nM (37) and 1354 nM (38)
5
0
demonstrate that a substitution with methoxy or thiomethyl is favourable for scaffold D.
Compound 37 possessed the highest inhibitory potency within this subset resulting in a
submicromolar IC50 value. Moreover, this compound was the only derivative in the subset
with a lower IC value than the 2-phenyl analogue (see 48). Furthermore, compounds 39 and
5
0
4
0 containing a 3-fluoro or 3-acetamido substituent, respectively, were synthesized. Resulting
IC values of 8120 nM and 28900 nM reveal a substantially lower inhibitory potency than
5
0
their phenyl analogues 52 (IC : 355 nM) and 53 (IC : 278 nM). As observed for scaffold C
5
0
50
with an amido linker, the derivatives based on scaffold D only reach between 55-75% of the
top value of Ko143.
1
3

ACCEPTED MANUSCRIPT
From the results of the Hoechst 33342 accumulation assay several findings can be highlighted:
first subset based on scaffold A with a thiophene or a pyrrole heterocycle at position 2 showed
some similarity regarding the activities to their phenyl analogues. However, both heterocycles
led to either similar or worse inhibitory potencies than their phenyl counterparts. The most
potent substituent in both cases was a 3-cyano group (see compounds 7 and 11).
Finally, we found that replacement of the aniline linker at position 4 by oxygen or sulpur in
all cases led to a considerable decrease in potency compared to their analogs. The same
demetrial effect on activity was observed by methylation of the aniline linker at position 4
2
(
scaffold B; R : NMe vs. NH). This emphasizes the importance of an anilino function with H-
donor properties at position 4 of the quinazoline scaffold for potent inhibitors.
Next, an amido-linker was investigated (scaffold C) instead of the amino-linker. Regarding
substituent positions considerable differences in inhibitory potencies were found, which were
in agreement with earlier findings pointing to disadvantage of an ortho-substitution and the
beneficial effect of nitro substituents [24]. Among the investigated compounds compound 31
with 3-nitro substituent (IC : 54.5 nM) showed the highest inhibitory potency. However, they
5
0
only reach between 59-70% of maximal inhibition in comparison to Ko143, which might be
due to a different mode of binding, which will be discussed in the following sections.
Furthermore, different derivatives using scaffold D were synthesized, lacking substitution
with phenyl at position 2. Except of compound 37 all investigated compounds showed
2
considerably lower potencies than their analogues based on scaffold B (R : NH). Also this
subset only reached an inhibition maximum between 55-75% compared to Ko143.
2
.2.2. Screening for ABCB1 and ABCC1 Inhibition. Compounds with high inhibitory
potency in the Hoechst 33342 accumulation assay were further investigated in a calcein AM
accumulation assay to determine their selectivity toward ABCG2. The assay was carried out
with the ABCB1 overexpressing cell line A2780adr as well as the ABCC1 overexpressing cell
1
4

ACCEPTED MANUSCRIPT
line H69 AR to cover the most prominent transport proteins in the context of MDR. Further
information about the execution of the assay is provided in the experimental section. The
screening was carried out with different compounds at a final concentration of 10 µM. As
positive control cyclosporine A (CsA) was used for both cell lines. Inhibition by the
compounds is given as percentage of inhibition with respect to the control CsA. A bar chart to
the results is depicted in Figure 4. In the case of inhibitory activity of more than 25 % a
dilution series was carried out to determine the IC of such compounds. The associated IC
50
50
values are given in Table 2.
a)
b)
1
5

ACCEPTED MANUSCRIPT
Figure 4. Inhibitory effect of screened compounds toward ABCB1 overexpressing cell line
A2780adr (a) and ABCC1 overexpressing cell line H69AR (b) in the calcein AM assay at a
concentration of 10 µM. Cyclosporine A (CsA) was used as positive control, indicating
complete inhibition. The inhibitory effect of each compound is expressed by the length of the
bars, representing the inhibition compared to the positive control in percent. For each
compound, three independent experiments were performed and the standard deviation is
expressed by error bars.
Table 2. Inhibitory activity of compounds exhibiting an inhibition of more than 25 % in
comparison to the reference cyclosporine A (CsA) in the calcein AM assay at a concentration
of 10 µM.
Calcein AM
ABCB1)
Scaffold IC ± SD [µM]
Calcein AM
(ABCC1)
(
1
2
a,b
a,c
Compd
8
R
R
IC ± SD [µM]
5
0
50
3,4-OMe
3-NHCOCH
3-CN
4-OMe
3,4-OMe
S
S
A
A
A
A
A
1.88 ± 0.71
14.7 ± 4.42
18.8 ± 0.87
8.03 ± 0.96
1.81 ± 0.21
1.17 ± 0.17
9.68 ± 0.26
n.t.
85.1 ± 7.9
n.t.
17.3 ± 1.9
3.53 ± 0.61
10
11
12
13
3
NH
NH
NH
d
Cyclosporine A
a
:
:
:
IC values were determined by at least three independent experiments.
50
b
The ABCB1 overexpressing cell line A2780adr was used.
The ABCC1 overexpressing cell line H69AR was used.
Cyclosporine A is used as reference for both assays.
c
d
:
n.t.: Not tested, due to low effect in the initial screening.
First compounds 5-13 were investigated for their inhibitory potency toward ABCB1. Among
the derivatives with a thiophene moiety at position 2 only compounds 8 and 10 exceed the 25 %
1
cut-off. Compound 8 contains 3,4-dimethoxy groups at R and yielded an IC of 1.88 µM
5
0
which is similar to the standard inhibitor CsA (IC : 1.17 µM). Substitution with a 3-
5
0
1
6

ACCEPTED MANUSCRIPT
acetamido group as in compound 10 resulted in an IC50 of 14.7 µM. This outcome is not
surprising, since earlier studies found that the presence of methoxy groups is favourable for a
potent ABCB1 inhibitor [24,25,26]. On the other hand, all pyrrole derivatives showed
inhibitory activities toward ABCB1 of more than 25 %. Compounds 11, 12 and 13, which
contain 3-cyano, 4-methoxy and 3,4-dimethoxy substituents, respectively, yielded IC values
5
0
of 18.8 µM, 8.03 µM and 1.81 µM in the given order. Compound 13 is nearly as potent as the
standard inhibitor CsA (IC : 1.17 µM). Again, the correlation of the amount of methoxy
5
0
groups present with the inhibitory activity toward ABCB1 is striking. Almost no activity was
found for compounds 31 and 32 containing an amido function at position 4. In accordance to
our previous studies, this also confirms the observation that nitro functions do not increase the
potency of compounds toward ABCB1 [24,25,26,27].
The calcein AM assay with ABCC1 overexpressing cells revealed among compounds 5-13,
with a five ring heterocycle at position 2, only three compounds that exceeded the 25 % cut-
1
off: compound 8 with a 3,4-dimethoxy substitution at R and thienyl ring at position 2 of the
quinazoline scaffold showed the highest inhibitory potency with an IC50 of 9.68 µM.
Compound 11 and 13, both with a pyrrole moiety at position 2, as well as 3-cyano and 3,4-
1
dimethoxy at R , respectively, only slightly exceed the 25 % cut-off.
2
.2.3. Determination of the Intrinsic Cytotoxicity. Intrinsic cytotoxicity of selected
compounds was determined with MDCK II BCRP overexpressing and parental cells. The
assay was carried out by incubating the cells with different dilutions of the test compound for
7
2 h. Subsequently, the cell viability was evaluated using 3-(4,5-dimethylthiazol-2-yl)-2,5-
diphenyltetrazolium bromide (MTT) which is reduced to its insoluble formazan species by
cellular reductases in living cells. Concentration of the dye can be measured spectrometrically
and correlated with the amount of living cells per well. More detailed information is provided
in the experimental section.
1
7

ACCEPTED MANUSCRIPT
Cytotoxicity of a compound is given as the concentration leading to 50 % growth inhibition
(
GI ). A control is established for 100 % cell viability using the same amount of DMSO (≤
50
1
%) and MeOH (≤ 1.8 %) as for the dilutions but without compound. Also, the state “0 %
cell viability” is induced by adding 10 % DMSO to one well. Table 3 summarizes the GI50
values for ABCG2 overexpressing cells as well as parental cells. Furthermore, the therapeutic
ratio (TR = GI /IC ) is given for the tested compounds to evaluate their possible use in an in
5
0
50
vivo application.
Table 3. Intrinsic Toxicity of Selected Compounds in ABCG2 Overexpressing MDCK II
BCRP and Parental Cells.
a
a
GI [µM]
GI [µM]
Therapeutic
5
0
50
1
2
Compd
R
R
Scaffold
BCRP
Parental
ratio (GI /IC )
5
0
50
5
6
7
3-NO
4-NO
3-CN
3-CN
H
2
S
S
S
A
A
A
A
C
C
C
D
13.8
83.2
58.9
38.8
71.7
1.35
0.904
13.8
12.6
1.36
14.8
105
63.1
55.0
128
4.50
1.80
8.9
12.5
2.10
46
248
331
249
169
25
9.7
21.9
56
2
11
29
31
32
37
NH
3-NO
4-NO
2
2
3-OMe
Ko143
gefitinib
0.80
a
:
Concentration accomplishing 50% of the maximal inhibition of cell proliferation with MDCK II BCRP and
parental cells, respectively. The data was obtained from at least two independent experiments as mean values.
First, compounds 5, 6 and 7 based on scaffold A with a thiophene moiety at position 2 were
investigated. Substitution was carried out with 3-nitro, 4-nitro and 3-cyano in the given order.
Most notably, a significant difference regarding the GI values of compounds 5 (GI : 13.8
5
0
50
µM) und 6 (GI : 83.2 µM) was found. Due to the fact that both compounds have a similar
5
0
IC value, the therapeutic ratio of 6 (TR: 248) is more than five-fold higher than that of 5 (TR:
5
0
1
8

ACCEPTED MANUSCRIPT
4
6) and still more than four-fold higher than the standard inhibitor Ko143 (TR: 55.5).
However, the highest TR was calculated for compound 7 (TR: 331), due to its low
cytotoxicity (GI : 58.9 µM) and high inhibitory activity (IC : 178 nM). Compound 11,
5
0
50
1
which also contains a 3-cyano substituent at R but a pyrrole moiety at position 2, possesses a
slightly higher cytotoxicity (GI : 38.8 µM) than compound 7, leading to a little lower TR of
5
0
1
2
7
49. The basic scaffold C with hydrogen at R (compound 29) showed a low cytotoxicity of
1.7 µM. However, compounds with the highest inhibitory potency in the whole test set,
namely 31 and 32, showed also the highest cytotoxic effects (31: GI : 1.35 µM; 32: GI :
5
0
50
0
.904 µM). Unfortunately, we found that the presence of nitro groups can contribute to
cytotoxic effects unless using a disubstitution of 3-hydroxy-4-nitro at the aromatic core at
position 4 [24]. Furthermore, the most potent compound among the derivatives based on
1
scaffold D was investigated: compound 37 with a 3-methoxy substituent at R yielded a GI50
of 13.8 µM which is in the range of Ko143. Interestingly it is 10-fold less toxic than the
structurally closely related TKI gefitinib, which also lacks substitution at position 2.
In summary, the five-membered heterocycles based on scaffold A exhibited the least
cytotoxicity among the investigated compounds together with the basic scaffold C present in
compound 29. Substitution with 3-cyano was preferable for low toxicities over 3-nitro (see
compounds 5, 7 and 11). However, a para nitro substituent as in compound 6 led to the lowest
cytotoxic effects. Highest toxicity was found with compounds 31 and 32 containing an amido
linker at position 4 and para or meta nitro substituents.
2
.2.4. Determination of the Ability to Reverse MDR. Reversal of the MDR induced by the
overexpression of ABCG2 was investigated for the 4 most potent compounds 7, 11, 31 and 32.
For this purpose the cytostatic agent mitoxantrone (MX) was used, since it is a substrate of
ABCG2. Highest cell viability is expected for the ABCG2 overexpressing cells in the absence
of any inhibitor. Likewise, a lower cell viability is obtained in the presence of a potent
1
9

ACCEPTED MANUSCRIPT
inhibitor due to the decreased efflux of the cytostatic MX. Concentration of MX was chosen
as 0.5 µM and different dilutions of the corresponding compound were added to the wells and
incubated for 72 h. Further information about the assay is provided in the experimental
section.
Cell viability of the corresponding compounds is given as a bar chart (Figure 5), where the
light grey bars represent the viability in the presence and the dark grey bars in the absence of
MX.
b)
a)
c)
d)
Figure 5. MDR reversal assay of compounds 7 (a), 11 (b), 31 (c) and 32 (d) demonstrating
their ability to reverse MDR toward the cytostatic mitoxantrone, in the ABCG2
overexpressing cell line MDCK II BCRP. The bars represent the cell viability at a given
modulator concentration in the presence (light grey) and absence (dark grey) of 0.5 µM
2
0

ACCEPTED MANUSCRIPT
mitoxantrone. Control shows viability of cells without modulator. The standard deviation is
expressed by error bars.
1
Compound 7, with a thienyl moiety at position 2 and 3-cyano at R of scaffold A showed no
cytotoxicity, which is illustrated by the constant cell viability in the absence of MX (Figure 5
a, dark bars) for all applied compound concentrations. Co-administered with MX (light bars) a
slow decrease of the cell viability was found for the two lowest compound concentration, but
attaining complete reversal of MDR at a concentration ≥ 1 µM. From the dose-effect curve
fitted from the cell viability an IC50 value of 45.3 nM was calculated. Compound 11
1
substituted with 3-cyano at R and pyrrole at position 2 of scaffold A exhibited no cytotoxic
effects as well (Figure 5 b, dark bars). Indeed, the decrease of the cell viability in the presence
of MX is more distinct at lower concentrations than with compound 7. Hence, the resulting
IC of 7.98 nM value is significantly lower. Complete reversal of MDR was achieved at a
5
0
compound concentration of about 3 µM. Moreover, compound 31 and 32 based on scaffold C
were investigated in the MTT efficacy assay. Compound 31 contains a 3-nitro substituent
1
while 32 has a 4-nitro substituent at R . According to the black bars (Figure 5, c, d)
representing the cell viability in the absence of MX at various compound concentrations, a
significant cytotoxic effect of both compounds can be observed. Compound 31 exhibits a
stronger cytotoxicity, which is in accordance with the results from the MTT cytotoxicity assay
presented above. The derived IC50 values of 47.3 nM (31) and 77.3 nM (32) from the cell
viability in presence of MX also reflect their difference in potency, which was observed in the
Hoechst 33342 accumulation assay (31: 54.5 nM; 32: 93.1 nM), very accurately. Both
compounds achieve a complete reversal of the MDR at a concentration of approximately ≥ 1
µM.
2
1

ACCEPTED MANUSCRIPT
2
.2.5. Investigation of ATPase activity. ATPase activity measurements were carried out with
High Five insect cell membranes, expressing ABCG2 protein after infection with baculovirus
containing ABCG2 cDNA.
We screened a representative selection of compounds of the most potent scaffolds found in
the Hoechst 33342 accumulation assay using three concentrations (0.1, 1 and 10 µM). The
vanadate-sensitive ATPase activity is given in comparison to the basal activity which was set
to 100 %. Table 4 shows the effect of the corresponding compounds on ATPase activity. In
case of activation or inhibition an additional Vmax value is given.
Table 4. Effect of compounds in the ATPase activity assay. All values are relative vanadate-
sensitive ATPase activities in relation to the basal activity, which is set to 100 %.
Vmax
1
2
Compound
R
R
Scaffold Effect on ATPase activity [% of control]
5
6
7
8
9
3-NO
4-NO
3-CN
3,4-OMe
3-F
2
S
S
S
S
S
A
A
A
A
A
A
A
activating
no effect
activating
activating
activating
activating
biphasic
no effect
no effect
inhibiting
no effect
no effect
no effect
activating
no effect
no effect
activating
activating
no effect
activating
activating
activating
activating
no effect
160
2
157
180
132
158
137
11
12
14
29
30
31
32
33
34
35
36
37
39
41
42
43
45
46
47
3-CN
4-OMe
NH
NH
H
C
C
C
C
D
D
D
D
D
D
B
B
B
B
B
B
2-NO
3-NO
4-NO
H
60
2
2
2
3-NO
4-OH
4-CN
-4-OH
150
2
3-OMe
3-F
H
157
140
a
NH
NH
NH
NH
NH
NH
3-NO
3-NO
4-OH
3-CN
4-CN
147
135
145
145
2
b
-4-OH
2
a
b
a
2
2

ACCEPTED MANUSCRIPT
4
5
5
9
0
2
4-OMe
3,4-OMe
3-F
NH
NH
NH
B
B
B
no effect
activating
activating
inhibiting
no effect
activating
140
142
14
b
b
Ko143
gefitinib
quercetin
b
177
a
:
:
Data taken from our previous studies [27].
Data taken from our previous studies [25].
b
The data was obtained from at least two independent experiments as mean values.
The effect of the compounds based on scaffold A on ATPase activity was mostly stimulating.
A significantly high stimulation of 180 % of control was found for 2-thienyl derivative 8 with
1
a 3,4-dimethoxy substitution at R , showing a comparable effect as the standard stimulator
quercetin (177 %). While all meta substituted derivatives were stimulating, the para
1
2
substituted ones showed a different effect on ATPase activity: compounds 6 (R : 4-NO , R : S)
2
1
2
and 12 (R : 4-OMe, R : NH) either led to no effect or showed a biphasic behavior,
respectively. Regarding compound 12 with a biphasic behavior, it is assumed that it binds to
an activating high affinity binding site and a low affinity inhibiting site. In the case of a
sufficient difference between the affinities of both binding sites a bell-shaped concentration
dependent curve is expected [29,30]. Although stimulation of ATPase activity is commonly
linked to a transport-activity induced by substrates, our previous work showed in detail that
this is not the general case [25,26,27]. For instance Hoechst 33342, a well-known substrate of
ABCG2, shows inhibition of ATPase activity with increasing compound concentrations (data
not shown), which has already been observed by Özvegy et al. [31].
2
For scaffold B (R : NH) we found similar effects as for the 2-phenyl-analogs: most
compounds had a stimulating effect on ATPase activity, in particular all meta substituted
1
derivatives at R . In contrast, two out of the three para substituted derivatives of this subset
showed no effect on ATPase activity (see compounds 47 and 49). Also, we did not observe
any effect on ATPase activity for the unsubstituted compound 41.
2
3

ACCEPTED MANUSCRIPT
Surprisingly, we found a different behavior for the subset based on scaffold C, with an amido-
linker at position 4 of the quinazoline scaffold. With exception of compound 30 all derivatives
had no effect on ATPase activity. Compound 30 on the other hand inhibited ATPase activity
with a Vmax of 60 %. Being the only compound with an ortho-functionalized aromatic moiety,
it can be promising to continue investigating this substitution pattern in terms of its impact on
ATPase activity. In our previous study of 2,4-substituted quinazolines we identified another
ortho-substituted compound (compound 30) as an inhibitor of ATPase activity [26]. However,
ortho-substituents were found to lead to poor ABCG2 inhibitors in the Hoechst 33342
accumulation assay, why we did not pursue this substitution pattern.
In the subset based on scaffold D, lacking the phenyl moiety at position 2 of the quinazoline
scaffold, we observed mixed results for ATPase activity. Again, meta-substituted compounds
as 37 and 39 led to stimulation of ATPase activity while para-substituted derivatives (see
compounds 35 and 36) and the unsubstituted derivative 33 had no effect on ATPase activity at
all.
This investigation could further confirm, that the impact of quinazoline derivatives on ATPase
activity is highly dependent on the substitution pattern at the anilino moiety at position 4.
Furthermore, we found that the type of linker (see amido vs amino) also plays a crucial role.
The gathered results can help to elucidate the complex mechanisms of the transport protein
ABCG2 even more.
2
.2.6. Investigation of Type of Interaction. Investigation of the type of inhibition of Hoechst
3342 was carried out for selected compounds 5, 6, 7, 11, 31, 32 and gefitinib. The assay was
3
carried out with the ABCG2 overexpressing MDCK II BCRP cell line using different
concentrations of Hoechst 33342 and compound. Thereby, the type of interaction of the
ABCG2 substrate Hoechst 33342 and the corresponding compound could be determined
applying the Lineweaver-Burk linearization and the direct linear plot according to Cornish-
2
4

ACCEPTED MANUSCRIPT
Bowden. Further details to the assay are provided in the experimental section. Exemplary
diagrams of compounds 5, 6, 31 and 32 are depicted in Figure 6.
a)
b)
4
3
2
1
conc. 5 [nM]
0
56
100
178
3160
0
5
10
15
1/S
c)
d)
5
4
3
2
1
conc. 6 [nM]
0
56
178
316
3160
0
10
/S
20
1
conc. 31 [nM]
e)
f)
1
3
5
1
0
4
2
2
6
00
3160
-
5
0
5
10
1/S
g)
h)
conc. 32 [nM]
6
0
5
1
1
3
3
6
4
2
00
78
16
160
2
5
5
10
1/S

ACCEPTED MANUSCRIPT
Figure 6. Lineweaver–Burk plot for compounds 5 (a), 6 (c), 31 (e) and 32 (g) at various
concentrations together with the ABCG2 substrate Hoechst 33342. Plot of the Michaelis-
Menten constant K (closed circle) and the maximum velocity V (closed square) derived
M
max
from the analysis according to Cornish-Bowden against the corresponding compound
concentration 5 (b), 6 (d), 31 (f) and 32 (h). Excluded values are depicted by the
corresponding open symbols.
Compounds 5, 6, 7, 11, 31 and 32 each yielded an intersection of the lines calculated from the
Lineweaver-Burk linearization in the second quadrant of the coordinate system. This is
interpreted as a non-competitive interaction with Hoechst 33342 but can also include
characteristics of a competitive interaction although not binding to the same pocket as
Hoechst 33342. Such “mixed-type” inhibitors can for instance trigger a conformational
change within the protein and thus influence the turnover rate of the substrate at the active site.
Due to the fact that the double reciprocal plot used in the Lineweaver-Burk linearization can
be susceptible to experimental error, the direct linear plot according to Cornish-Bowden was
additionally applied out to confirm the results of the Lineweaver-Burk plot and to get more
insights into the underlying mechanism. For all compounds a clear non-competitive behaviour
was confirmed with the Cornish-Bowden plot, illustrated by an approximately constant KM
value and a decreasing Vmax with increasing compound concentration. The tyrosine kinase
inhbitor gefitinib and the reference inhibitor Ko143 yielded an intersection in the third
quadrant the Lineweaver-Burk plot which is also characteristic for a non-competitive
interaction. This was additionally confirmed by the Cornish-Bowden plot resulting in constant
K values and decreasing V values with increasing compound concentration.
M
max
2
6

ACCEPTED MANUSCRIPT
All investigated compounds showed a non-competitive interaction of the mixed type with
Hoechst 33342. This means that their binding-pocket is different from Hoechst 33342 but is
likely to have an influence via allosteric interactions. Due to the fact that there is still limited
knowledge about the binding sites of ABCG2, this inhibitory kinetic study contributes to a
better understanding of the interaction and binding of inhibitors toward ABCG2 [32,33].
2
.2.7. 5D3 antibody binding assay. Investigation of the conformational change of ABCG2 in
the presence of an inhibitor was carried out with the conformational sensitive monoclonal
TM
antibody PerCP-Cy 5.5 Mouse Anti-Human CD338. This conformationally sensitive
antibody binds specifically to the CD338 epitope of ABCG2 and contains a fluorescent dye
which can be excited with a blue laser (488 nm). The study was carried out with the PLB-985
ABCG2 overexpressing cell line and the fluorescence of the bound antibody was measured
with a FACSCalibur cytometer. Here, the four most potent compounds (7, 11, 31, 32), the
tyrosine kinase inhibitor gefitinib and the ABCG2 substrate Hoechst 33342 were selected. All
compounds were measured at a concentration of 10 µM and the obtained fluorescence with
Ko143 set as reference, representing 100 % labelling of ABCG2 with the antibody. The
degree of immunostaining by each compound is presented as bar-chart in Figure 7. A
histogram showing the highest and the lowest 5D3-shift in the assay is presented in Figure 8.
2
7

ACCEPTED MANUSCRIPT
100
50
0
3
inib
n
2
7
1
1
2
4
i
4
3
1
3
3
t
1
e
o
efit
c
3
K
r
3
e
t
G
u
s
h
Q
c
e
o
H
Figure 7. 5D3 immunoreactivity modulation of ABCG2 by various compounds at a
concentration of 10 µM. Fluorescence detected by the 5D3-labeling of ABCG2 in the
presence of 10 µM Ko143 was set to 100 % and the fluorescence measured in the absence of
any compound taken as 0 %.
a)
b)
Figure 8. Histogram of the measured fluorescence at the FL3-H detector (X-axis) and the
cell-count gated according to the fluorescence. Depicted is the fluorescence of the isotype-
control (dotted curve) as well as of 5D3 antibody in the absence of a compound (dashed curve)
and in the presence of a compound (continuous curve).Compounds with the highest and
lowest 5D3 shifts: Ko143 (a) and 32 (b) at a concentration of 10 µM.
2
8