Targeting DNA Repair in Tumor Cells via Inhibition of ERCC1-XPF

Article abstract of DOI:10.1021/acs.jmedchem.9b00326

The ERCC1-XPF heterodimer is a 5′-3′ structure-specific endonuclease, which plays an essential role in several DNA repair pathways in mammalian cells. ERCC1-XPF is primarily involved in the repair of chemically induced helix-distorting and bulky DNA lesions, such as cyclobutane pyrimidine dimers (CPDs), and DNA interstrand cross-links. Inhibition of ERCC1-XPF has been shown to potentiate cytotoxicity of platinum-based drugs and cyclophosphamide in cancer cells. In this study, the previously described ERCC1-XPF inhibitor 4-((6-chloro-2-methoxyacridin-9-yl)amino)-2-((4-methylpiperazin-1-yl)methyl)phenol (compound 1) was used as a reference compound. Following the outcome of docking-based virtual screening (VS), we synthesized seven novel derivatives of 1 that were identified in silico as being likely to have high binding affinity for the ERCC1-XPF heterodimerization interface by interacting with the XPF double helix-hairpin-helix (HhH2) domain. Two of the new compounds, 4-((6-chloro-2-methoxyacridin-9-yl)amino)-2-((4-cyclohexylpiperazin-1-yl)methyl)phenol (compound 3) and 4-((6-chloro-2-methoxyacridin-9-yl)amino)-2-((4-(2-(dimethylamino)ethyl) piperazin-1-yl) methyl) phenol (compound 4), were shown to be potent inhibitors of ERCC1-XPF activity in vitro. Compound 4 showed significant inhibition of the removal of CPDs in UV-irradiated cells and the capacity to sensitize colorectal cancer cells to UV radiation and cyclophosphamide.

Full text of DOI:10.1021/acs.jmedchem.9b00326

Subscriber access provided by KEAN UNIV
Article
Targeting DNA Repair in Tumor Cells via Inhibition of ERCC1-XPF
Ahmed Elmenoufy, Francesco Gentile, David Jay, Feridoun Karimi-
Busheri, Xiaoyan Yang, Olivier M Soueidan, Claudia Weilbeer, Rajam S
Mani, Khaled H. Barakat, Jack Tuszynski, Michael Weinfeld, and F. G. West
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.9b00326 • Publication Date (Web): 01 Aug 2019
Downloaded from pubs.acs.org on August 2, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 38
Journal of Medicinal Chemistry
1
2
3
4
Targeting DNA Repair in Tumor Cells via Inhibition of ERCC1-XPF
5
6
7
8
9
Ahmed H. Elmenoufy1,2, Francesco Gentile3, David Jay4, Feridoun Karimi-Busheri4, Xiaoyan Yang4,
Olivier M. Soueidan1, Claudia Weilbeer1, Rajam S. Mani4, Khaled H. Barakat5, Jack A. Tuszynski3,4,
Michael Weinfeld4 and Frederick G. West1*
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
1Department of Chemistry, University of Alberta, Edmonton, Alberta, Canada T6G 2G2
2Department of Pharmaceutical Chemistry, College of Pharmacy, Misr University for Science and
Technology, 6th of October City, Egypt, P.O. Box: 77
3Department of Physics, University of Alberta, Edmonton, Alberta, Canada T6G 2E1
4Department of Oncology, Cross Cancer Institute, University of Alberta, Edmonton, Alberta, Canada
T6G 1Z2
5Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada
T6G 2H1
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Abstract
The ERCC1-XPF heterodimer is a 5´-3´ structure-specific endonuclease, which plays an essential role in
several DNA repair pathways in mammalian cells. ERCC1-XPF is primarily involved in the repair of
chemically-induced helix-distorting and bulky DNA lesions, such as cyclobutane pyrimidine dimers
(CPDs), and DNA interstrand crosslinks. Inhibition of ERCC1-XPF has been shown to potentiate
cytotoxicity of platinum-based drugs and cyclophosphamide in cancer cells. In this study, the previously
described ERCC1-XPF inhibitor 4-((6-chloro-2-methoxyacridin-9-yl)amino)-2-((4-methylpiperazin-1-
yl)methyl)phenol (compound 1) was used as a reference compound. Following the outcome of docking-
based virtual screening (VS), we synthesized seven novel derivatives of 1 that were identified in silico
1
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 38
1
2
3
4
5
6
7
8
9
as being likely to have high binding affinity for the ERCC1-XPF heterodimerization interface by
interacting with the XPF double helix–hairpin–helix (HhH2) domain. Two of the new compounds, 4-((6-
chloro-2-methoxyacridin-9-yl)amino)-2-((4-cyclohexylpiperazin-1-yl)methyl)phenol (compound 3) and
4-((6-chloro-2-methoxyacridin-9-yl)amino)-2-((4-(2-(dimethylamino)ethyl) piperazin-1-yl) methyl)
phenol (compound 4), were shown to be potent inhibitors of ERCC1-XPF activity in vitro. Compound 4
showed a significant inhibition of the removal of CPDs in UV-irradiated cells and the capacity to sensitize
colorectal cancer cells to UV radiation and cyclophosphamide.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Introduction
The ERCC1-XPF heterodimer is a critical DNA repair endonuclease. It plays a pivotal role in nucleotide
excision repair (NER) of bulky adducts and helix-distorting DNA lesions such as UV-induced
pyrimidine-(6,4)-pyrimidone photoproducts (6-4PPs) and CPDs 1-4. ERCC1-XPF is also involved in
DNA interstrand crosslink (ICL) repair 5 in cells treated with platinum-based and other chemotherapeutic
agents such as cyclophosphamide and mitomycin C (MMC) 6. In a recent study, 72% of colorectal cancer
patients showed positive ERCC1 protein expression which may be a useful marker for colorectal cancer
patients 7. There is also evidence that ERCC1-XPF participates in DNA double-strand break (DSB) repair
8, 9. It thus contributes significantly to the response of cancer cells to a range of DNA-damaging
chemotherapeutic agents and radiotherapy. In the ERCC1–XPF complex structure, ERCC1 is considered
to be catalytically inactive but rather regulates DNA–protein and protein–protein interactions, whereas
the endonuclease activity is provided by XPF, which also contains an inactive helicase-like motif that is
likely to be involved in protein–protein interactions and DNA binding 10, 11
.
ERCC1-XPF is an attractive target for drug design due to the availability of experimental structures and
the presence of multiple sites that can be inhibited with small molecules to stop the activity of the
12
endonuclease. Our earlier drug design studies focused on the XPA-ERCC1 interaction site and the
2
ACS Paragon Plus Environment

Page 3 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
XPF active site 13, fostering the use of computer-aided drug design to develop DNA repair inhibitors 14.
However, due to the specificity of the ERCC1-XPF interaction and its general involvement in all the
ERCC1-XPF mediated repair processes, the dimerization interface is perhaps the most promising domain
8
9
4
to target pharmacologically . Dimerization and localization of ERCC1 and XPF is essential for the
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
enzyme’s stability and endonuclease activity 15. The dimerization of C-terminal regions of ERCC1 and
XPF is the key interaction to form a stable heterodimer. C-terminal regions dimerize through the
interaction of their double helix–hairpin–helix (HhH2) motifs 16, 17. It is thought that XPF acts as a
scaffold for ERCC1 during protein folding, and ERCC1 may exhibit improper folding in vitro in the
17
absence of XPF . It was demonstrated that without dimerization the activity of ERCC1-XPF was
abolished because neither protein was stable, and therefore they were rapidly degraded 18, 19. As a result,
development of small molecule inhibitors that can disrupt the HhH2domain interactions between ERCC1
and XPF would be expected to sensitize cancer cells to chemotherapeutic treatments whose DNA-
4
damaging effects are repaired by ERCC1–XPF-dependent pathways . Moreover, rational drug design
methodology can be employed due to the availability of multiple experimental structures of the dimerized
HhH2 domains (for example, PDB code 2A1J and 1Z00) 19.
Recently, McNeil et al. employed an in-silico screening approach targeting three sites on the XPF HhH2
domain to identify possible inhibitors for the dimer. They discovered a small molecule able to inhibit the
NER activity in melanoma cells and slightly sensitize them to cisplatin treatment. However, the reported
20
Kd and IC50 values for this compound were suboptimal, in the medium-high µM range . In addition,
Chapman et al. 21, 22, and Arora et al. 23 identified and optimized different small molecules targeting the
active site on the XPF nuclease domain. These efforts resulted in several endonuclease inhibitors with
IC50 in the nanomolar range, able to diminish NER activity and enhance the cytotoxicity of platinum-
based drugs in cancer cells. Although the specificity of these inhibitors to the ERCC1-XPF endonuclease
was assessed in some cases, similarities among active sites of various endonucleases could produce off-
3
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 38
1
2
3
4
5
6
7
8
9
target interactions from these compounds; the lack of structural insights of the ligand-protein complexes
in this case also limits further development of this series. Most recently, Yang et al. proposed the cellular
delivery of therapeutic peptides mimicking the ERCC1 HhH2 domain (residues 220-297) as a promising
alternative strategy to inhibit NER activity and sensitize cancer cells to DNA-damaging agents 24.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Development of small molecule inhibitors of the HhH2 domain interaction would be expected to sensitize
cells to therapies whose DNA-damaging effects are repaired by ERCC1–XPF-dependent pathways.
Jordheim et al, focused on developing small molecule inhibitors of the NER pathway acting through the
inhibition of ERCC1-XPF heterodimerization and reported that compound 1 (also called F06, NSC-
130813 or NERI02) interacted with XPF, repressed the interaction between ERCC1 and XPF in vitro
25
and sensitized cancer cells to MMC and cisplatin . This affirms that targeting this protein-protein
interaction can enhance the cytotoxic activity of crosslinking agents such as cisplatin. The initial hit, 1,
arose from a virtual screening (VS) of a large compound library, and the study also provided a
characterization of the XPF binding pocket and the binding mode of the compound to it 25. Compound 1
was predicted to interfere with the heterodimerization of ERCC1 and XPF, a necessary step to attain
DNA repair activity.
Preliminary in-vitro assays confirmed that 1 shows promising inhibitory activity and acts synergistically
with cisplatin. However, the activity of 1 is suboptimal in terms of clinical properties including potency
25
and safety, and a derivatization strategy, suggested by Jordheim et al , was adapted to optimize the
action of the compound. Also, previous efforts on developing inhibitors for ERCC1-XPF activity were
carried out on the truncated version of the heterodimer, and not the full-length protein 2, 25-27. It is reported
that the full length ERCC1–XPF protein was 15-fold more active than the truncated form under standard
reaction conditions 2, 19. To the best of our knowledge, we are the first group to report inhibitors of the
activity of the full-length protein. Using 1 as a reference hit, we employed computer-aided drug design
techniques such as electrostatic mapping of the compound 1 binding pocket, molecular docking,
4
ACS Paragon Plus Environment

Page 5 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pharmacophore modeling and molecular dynamics (MD)-based rescoring to rank compound 1
derivatives based on their predicted binding affinities to the XPF domain. The reference compound
(compound 1) and top hits (compounds 2-7) were synthesized and tested for their ability to inhibit the
in-vitro endonuclease activity of the full length ERCC1-XPF protein, and the most active compound was
further assessed as an inhibitor of the repair of UV-induced thymidine dimers in colorectal cancer cells,
as well as a sensitizing agent to not only to UV radiation but also to cyclophosphamide which has been
used to significantly reduce proliferation of metastatic colorectal cancer 28.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Results and Discussion
Computer-aided drug design of compound 1 analogues
The first step was to investigate the binding mode of the hit compound, 1, to the pocket on the XPF C-
terminus, followed by functionalization and extension of the piperazine ring in compound 1 : 1) provide
potential key interactions via hydrogen bond formation or hydrophobic interactions, and 2) optimize the
physicochemical properties and binding affinity for better potency and reduced toxicity of the compound
1 based compounds.
The binding energy for the best docked conformation of 1 to the XPF structures was -10.23 kcal/mol, as
calculated by the Autodock scoring function. The compound showed a high shape complementarity with
the XPF pocket, which interacts with the F293 residue of ERCC1 in the dimerized complex (Figure S2A).
Three ligand-receptor hydrogen bonds were observed in the docked pose, namely with the side chain of
E829, the backbone of N834 and the backbone of K860. In addition, the hydrophobic core of 1,
comprised of three aromatic rings, was positioned within a hydrophobic zone constituted by XPF residues
Y833, Q838, M856 and H857 (Figure S2B). Also, the binding mode of the ligand was in accordance
with the spatial distribution of the affinity maps calculated in Molecular Operating Environment 2015
(MOE2015) (Figure S2C).
5
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 38
1
2
3
4
5
6
7
8
9
To correctly place the scaffold common to all the analogues within the binding site, a pharmacophore
model based on the binding pose of 1 was built using MOE. This model included six features, namely
four aromatic moieties, one hydrogen bond donor, and one hydrogen bond acceptor features (Figure S2D
and Table S1 in Supporting Information). Upon generation of such features using MOE, we manually
adjusted their radii consistently with the spatial distribution of the affinity maps. The total number of
molecular structures that were docked was fifty-seven, accounting for all the different states generated
during the preparation step. After the docking was performed, just the top scored pose for each analog
was retained. The results were ranked according to their Generalized Born Volume Integral/Weighted
Surface Area (GBVI/WSA) score, as calculated by MOE. Then, we performed the MD simulations of
the complexes, and rescored the analogues using the Molecular Mechanics-Generalized Born Surface
Area (MM/GBSA) method. A subset of the top ranked hits was selected for chemical synthesis.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The results of the docking-based VS for this subset are reported in Table S2. The first two hits
(compounds 3 and 4) were particularly interesting because of their binding energies in the range of the
lead compound, reasonable logP values and conserved ligand efficiencies with respect to the hit structure.
The decomposition of the binding energies among the residues of the binding pocket revealed a similar
pattern of ligand-receptor interactions for the compound 3, 4 and 1, although some differences are
noticeable for compound 4.
The residues that contribute consistently to the binding of compound 1 analogues were identified as
Y833, M856, H857 and V859, where van der Waals interactions dominated the binding. Electrostatic
interactions were divided in highly-favorable with acidic residues (D823, E825, E829, E831, D839,
E864), and highly-unfavorable with basic residues (K832, K843, K850, R853, H857, H858, K860), due
in part to the positive protonation state of the piperazine ring of the compounds (Figure 1A). Noteworthy,
both contributions were generally higher for compound 4, when compared with compound 1 and 3. This
may be due to the addition of another positive R-substituent moiety in compound 4, differently from and
6
ACS Paragon Plus Environment

Page 7 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
compound 1 and 3 where hydrophobic groups were present as R-substituents. A further confirmation of
the different nature of the binding was found when analyzing the different contributions of the total
binding energy. Indeed, while 1 and compound 3 showed favorable electrostatic interactions, a highly-
unfavorable value was observed for compound 4. An opposite trend was observed for the polar solvation
energy, highly-favorable for compound 4, and highly-unfavorable for compound 1 and 3. Van der Waals
and non-polar solvation contributions were similar for all the three compounds (Table S3). A detailed
analysis of the binding mode of compound 4 is shown in Figure 1B. The conserved core structure of the
analogue interacted similarly to 1 with the residues constituting the binding site, although the backbone
of V859 replaces K860 as hydrogen bond partner during the simulation. In addition, a network of water
molecules surrounding the binding site and interacting with the exposed R-substituent of compound 4
was observed during the 2-ns simulation (Figure 1C). These water-ligand interactions may be responsible
for the favorable polar solvation contribution of the binding energy of compound 4. Charge-charge
interactions with the E829 and E831 residues were maintained in terms of average magnitude, when
compared with 1. Based on our simulations and because the branch containing the piperazine ring in 1 is
exposed to the solvent and cannot be accommodated in any additional cleft, we believe that adding a
hydrophilic extension such as for compound 4 strongly improves the ligand-solvent interactions.
Consistent with the importance of limiting the hydrophobicity of the piperazine ring extension,
compound 1 analogues carrying aromatic rings in their R-substituents showed reduced calculated binding
affinities (Table S2) and displayed inactivity in our incision assay (Figure 2). Importantly, charge-charge
interactions with E829 and E831 residues of XPF were maintained for compound 4, when compared with
1.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
7
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 38
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 1. Analysis of the binding mode of the top hits. A) Per-residue decomposition of ligand-receptor
binding energy. Compound 1 and 3 had similar patterns, while compound 4 showed a different pattern
of interaction energies. Blue cells indicate favorable interactions, while red cells indicate unfavorable
interactions. Refer to the text for more details. B) Lowest potential energy snapshot extracted from the
MD simulation of compound 4 bound to XPF. Hydrogen bonds are colored in purple. The highest pocket
8
ACS Paragon Plus Environment

Page 9 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
contributors in terms of total interaction energy are labelled. C) Interaction diagram for compound 4 and
surrounding residues and water molecules. The non-bonded interactions established with water
molecules was responsible for a highly-favorable polar solvation contribution to the binding energy. The
diagram refers to the lowest potential energy snapshot extracted from the MD simulation. Original PDB
structure for XPF: 1Z00.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Synthesis of compound 1 based analogues
Synthesis of compounds 1-8 was achieved through a one-pot sequential addition reaction in 3 steps as
shown in Figure S1A. Mannich reaction of p-acetamidophenol with formaldehyde and appropriate
secondary amine in 2-propanol was carried out under reflux for 12 h. The solvent and excess of the
unreacted formaldehyde from the resulting mixture was removed under vacuum and without isolating
the compound, the resulting viscous residue was treated with 6 M HCl to deacetylate the acetamido group
to furnish the primary amine as depicted in Figure S1. Afterwards, an equimolar amount of 6,9-dichloro-
2-methoxyacridine was added, affording after heating compounds 1-8 in moderate to good yields after
isolation. The sequence is general, facile and reproducible. All synthesized compounds 1-8 were
1
13
characterized by H NMR, C NMR, HRMS, IR and the purity of the most active compound 4 was
determined by HPLC (≥98% purity) as shown in page S32.
According to the in-silico screening, seven compounds were selected to be synthesized among the top
hits due to their promising ligand efficiencies and binding energy values in comparison to 1.
Derivatization focused primarily on extension of the piperazine ring with different functionalities (i.e
alicyclic (compound 3), substituted aromatics with electron donating (EDG) or electron withdrawing
(EWG) groups (compounds 2, 5, 6 and 7), and an aliphatic chain installed with hydrogen bond acceptor
(HBA) atom (compound 4) as shown in Table 1. Compound 8 was synthesized as a control to investigate
whether the piperazine ring is crucial for the activity of compounds.
9
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 38
1
2
3
Table 1.Compound 1 analogues functionalized with different R substituents.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 8 (control)
Compound
1 (reference)
2
R
3
4
10
ACS Paragon Plus Environment

Page 11 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
5
6
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
7
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Inhibition of ERCC1-XPF endonuclease activity
We used an in-vitro real-time fluorescence-based assay to assess the inhibitory effect of the synthesized
compounds (except compound 2 due to its instability over time) on ERCC1-XPF endonuclease activity.
29
This assay has been previously described , and utilizes a stem-loop substrate (composed of a 10mer-
duplex stem and a 20mer-oligodT single-strand loop), labelled on the 5′-terminus with 6-FAM and 3′
with the quencher dabcyl. Upon ERCC1-XPF cleavage in the 10mer-duplex region an 8-base 5′ FAM-
labelled product is released resulting in an increased fluorescent signal (Figure 2A, Control). The other
tracings in Figure 2A indicate to what extent the different compounds (10 µM each) could inhibit scission
of the substrate.
11
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 38
1
2
3
4
5
6
7
8
9
As shown here, three compounds exhibited a marked capacity to inhibit the nuclease activity of ERCC1-
XPF, i.e. compounds 1, 3 and 4. For these three compounds, different concentrations of the drugs were
plotted against the initial velocity (Vo, change in relative fluorescence units (RFU)/time) of the enzyme.
Figure 2B (inset) shows an example of data obtained with compound 4. Half-maximum inhibitory
concentrations (IC50) for these compounds were estimated from at least three different experiments for
every compound (Table S4).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 2. In-vitro inhibition of ERCC1-XPF endonuclease activity. A microplate fluorescent assay was
used to measure inhibition of ERCC1-XPF endonuclease activity by the different compounds. Incubation
of the DNA stem-loop substrate with ERCC1-XPF resulted in the release of a fluorescent 8-base
fragment. A) Shows the increase in fluorescence (FRU) with time. A representative tracing of the effect
12
ACS Paragon Plus Environment

Page 13 of 38
1
Journal of Medicinal Chemistry
2
3
of the different compounds (8 µM each) on the incision activity is shown. B) (inset) shows a
4
5
representative plot of enzyme rate (∆RFU/time) vs compound 4 concentration.
6
7
8
9
Since acridine compounds are known to bind nucleic acids, the possibility existed that a disruption of the
structure of the stem-loop substrate by compound 4 could have been responsible for the inhibition
observed in the ERCC1-XPF kinetic assays. To rule out this possibility the experiment shown in Figure
S4 was carried out. Pre-incubation of ERCC1-XPF with compound 4 (3 μM) in the reaction medium
resulted in a time-dependent loss of enzyme activity. The reactions were started either with addition of
the substrate (T0) or with substrate that had been pre-incubated in the reaction medium in the presence
of compound 4 (3 μM) for 12 minutes (T12). The data indicate that the time-dependent inactivation of
ERCC1-XPF by compound 4 was the consequence of the interaction of the drug with the enzyme, and
not altered by addition of substrate that had been pre-incubated with the inhibitor. The data in Figure
S4A were obtained with a concentration of compound 4 approximately 9 times higher than the observed
IC50 value of the compound and in the presence of 10% DMSO. When the same test was carried out in
the presence of 5% DMSO, a similar time-dependent inhibition was observed and the substrate exhibited
comparable behavior (Figure S4B). (The IC50 values reported in the present study were obtained by
incubating ERCCI-XPF in the presence of the indicated concentrations of the drug for 30 minutes in the
reaction medium containing 5% DMSO.)
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Inhibitor binding to ERCC1-XPF
Intrinsic fluorescence spectroscopy (of the protein tryptophan residues) was utilized to study the binding
affinity of compound 4 (active compound) and compound 5 (non-active compound) to ERCC1-XPF
(Figure 3). Addition of 2 µM of compound 5 had no significant effect on protein fluorescence and the
observed fluorescence quenching at 330 nm was only 3 ± 1%, thus providing no evidence of any
interaction. In contrast, addition of compound 4 (2 µM) induced nearly 23 ± 2% quenching of protein
13
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 38
1
2
3
4
5
6
7
8
9
fluorescence at 330 nm, clearly indicating interaction of compound 4 with the ERCC1-XPF complex as
depicted in Figure 3B. Binding affinity (in terms of dissociation constant, Kd) of compound 4 for
ERCC1-XPF complex was determined by following fluorescence quenching (a measure of ligand
binding) as a function of ligand concentration. A representative plot of relative fluorescence intensities
versus the concentration of compound 4 is shown in Figure 3C (inset). Nonlinear regression analysis
(GraphPad Prism Software, San Diego, CA) of the binding data was carried out as described in our earlier
paper (41) and revealed a unimodal binding with a Kd value of 100 ± 5 nM as also shown in Table S4.
Interestingly, compound 4 has has a high binding affinity fo ERCC1-XPF that is almost 20-fold greater
than its binding affinity to the DNA stem loop substrate used in the assay which has a unimodal binding
with a Kd value of 2.1 ± 0.1 µM. This disparity in binding affinity strongly suggests that inhibition by 4
results from its interaction with ERCC1-XPF rather than with the DNA substrate
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
14
ACS Paragon Plus Environment

Page 15 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 3. Determination of the affinity (Kd) between compound 4 and the ERCC1-XPF complex.
Quenching of the intrinsic protein fluorescence of the tryptophan residues was used to monitor the
interaction between the compounds and the protein. A) Fluorescence of ERCC1-XPF (13 nM and 20
nM) incubated with compound 5 (negative control), and B) compound 4 (active inhibitor) respectively.
C) Unimodal binding pattern and the binding affinity of compound 4 with ERCC1- XPF (70 nM). The
protein was excited at 295 nm and fluorescence intensity was monitored at 330 nm (see inset). The
fraction bound (i.e. relative fluorescence) intensity versus ligand concentration is plotted. D) Binding of
Compound 4 to DNA stem-loop substrate of ERCC1-XPF complex. The substrate of ERCC1-XPF was
excited at 490 nm and the fluorescence intensity at 520 nm was monitored as a function of Compound 4
15
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 38
1
2
3
4
5
concentration. A unimodal binding pattern was observed and the binding affinity of compound 4 to
ERCC1-XPF substrate was 2.1 ± 0.1 µM.
6
7
8
9
In an effort to identify a direct interaction between compound 4 and ERCC1-XPF we decided to utilize
microscale thermophoresis (MST) technique. We employed dERCC1 (ERCC1 codons 96-297) and
dXPF (XPF codons 667-916) for this analysis because these peptides contain the subunit interaction
domains that compound 4 was specifically designed to disrupt. Considering that dERCC1 is produced in
low yields in E. coli and dXPF, although highly expressed, is mainly generated as an insoluble fraction
in the absence of detergents 2. MST allows for the detection of interactions and changes in the molecular
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
properties of biomolecules at low nanomolar concentrations 30-32
.
The expectation was that compound 4 would induce a larger change in the thermophoretic properties of
the truncated dimer as compared to each of the individual peptides provided the drug interacted with the
enzyme according to the proposed mechanism.
Figure S5 shows the thermophoretic property of the fluorescently labeled proteins in the presence or
absence of compound 4. Initiation of the temperature gradient generated a depletion of the fluorescent
signal in the case of dXPF (grey line). In the case of dERCC1, after an initial depletion, the temperature
gradient induced a general accumulation of the fluorescent signal (light blue); a behavior that has been
30
observed before for other molecules . The dERCC1-dXPF dimer behaved in an intermediate fashion
(dark blue). Addition of compound 4 induced a reproducible minimal change in the dERCC1 (orange)
and dXPF (yellow) signals. Importantly, compound 4 changed more dramatically the signal of the
dERCC1-dXPF dimer (green vs dark blue), indicating that compound 4 interacts with ERCC1-XPF in
the subunit interaction domain and alters the properties of this region.
Inhibition of cellular repair of cyclobutane pyrimidine dimers
Significant inhibition of NER determined by the removal of UV-induced CPDs compared with control
cells over 24 hours is depicted in Figure 4. Immunofluorescent detection of CPDs after the exposure of
16
ACS Paragon Plus Environment

Page 17 of 38
Journal of Medicinal Chemistry
1
2
3
HCT-116 colorectal cells to 8 J/m2 treated with compound 4 showed a significant inhibition of removal
4
5
33
of CPDs compared with control cells over 24 hours. Based on our results and others , approximately
6
7
8
9
80% of CPDs are removed over 24 hours after UV exposure of HCT-116 cells, but this was reduced to
approximately 60% in the presence of compound 4. On the other hand, compound 5, which demonstrated
very limited inhibition in vitro, failed to inhibit the cellular removal of CPDs. Figure 4B shows the
relative quantification of CPDs after treatment with active (4) and inactive (5) compounds.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 4. Inhibition of cellular NER. A) Immunofluorescence images of the UV-based assay for
detecting CPDs in HCT-116 cells treated with compounds 4 and 5. B) Normalized fluorescence intensity
of the treated cells (±S.E.M) based on quantitation of fluorescence from 50 randomly selected cells per
treatment. All the values obtained for treatment with compound 4 from 4-24 h post-irradiation were
significantly different (p < 0.005, Student’s t-test) from the values obtained with the control irradiated
cells.
Increased sensitization to UV radiation and cyclophosphamide
We also evaluated the effectiveness of compound 4 (1 and 2 µM non-cytotoxic doses were chosen
according the clonogenic survival assay of compound 4 as shown in Figure S3) to sensitize cells to UV
17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 38
1
2
3
irradiation based on clonogenic survival as shown in Figure 5A. In accord with the repair data, 2 µM
4
5
compound 4 significantly reduced survival of the UV irradiated cells.
6
7
8
9
We then examined the combined effect of compound 4 (1 and 2 µM) on cellular survival following
exposure to the DNA interstrand crosslinking agent cyclophosphamide using HCT 116 (WT) cells and
XPF knockout cell lines. The survival curves shown in Figure 5B indicate that, at a concentration of 2
µM, compound 4 significantly sensitized the cells to cyclophosphamide (starting at 50 µM) and no
colonies were detected on the plates with a cyclophosphamide concentration of 300 µM.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 5. Sensitization of cells to UV and cyclophosphamide. A) Survival of HCT-116 cells exposed to
increasing doses of 254 nm UV radiation and treated with 1 and 2 µM compound 4 determined by the
clonogenic survival assay. B) Survival of HCT-116 cells exposed to increasing doses of
cyclophosphamide and treated with 1 and 2 μM compound 4 determined by the clonogenic survival
assay. The dashed line indicates that the no colonies were observed after treatment with 300 μM
cyclophosphamide + 2 μM compound 4. The survival curves (±S.E.M) are based on three independent
sets of determinations. All the values obtained with cells treated with 2 μM compound 4 were
significantly different (p < 0.005, Student’s t-test) from the values obtained with the control cells not
treated with compound 4. C) Human XPF gene knock out confirmation on two positive clone 1-2-1 and
8-1-2 by Western blot analysis using monoclonal antibodies against XPF and Mre11, as control proteins.
Pharmacokinetic properties of Compound 4 and compound 1
18
ACS Paragon Plus Environment

Page 19 of 38
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Parental compound 1 of our lead compound 4 was submitted to WuXi AppTec (Shanghai) Co.44 in China
to further investigate the ADME profile of the two compounds and compare their results to know whether
compound 4 is pharmacokinetically more superior than compound 1. Compound 4 showed a lower LogD
(2.86) than 1 (3.86) at pH 7.4 (Table S5). Metabolism of compound 4, as measured by human liver
microsome indicated that it has a metabolism in the mid-range, while 1 has a more rapid metabolism.
Both compounds display an efflux ratio of higher than 2 (compound 4 = 8.92 and compound 1 = 11.18).
Analysis of five CYP 450 tested enzymes indicates that neither of the compounds has an IC50 lower than
1.00 µM therefore they are not potent inhibitors of these enzymes. Compound 4, however, moderately
inhibits CYP1A2 with an IC50 of 6.40 µM. Compound 1 has the same moderate inhibitory effect on
CYP2D6 and CYP3A4-M with an IC50 between 1 to 10 µM. Compound 4 acts as a weak inhibitor of
CYP2D6 and CYP3A4-M while 1 is a weak inhibitor of three CYP 450 enzymes (CYP1A2, CYP2C9,
and CYP2C19) with an IC50 between 10 to 50 µM. Both compounds have similar behavior in terms of
permeability and binding to serum proteins as shown in Table S5.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
To conclude, our comparison of the ADME results between 1, the parental compound, and compound 4
indicated that the latter has more favourable pharmacological properties, particularly a lower LogD than
1, which means that it has a lower lipophilicity, and greater metabolic stability.
Conclusion
The ERCC1-XPF heterodimer is a structure-specific endonuclease, which is required especially for NER
and ICL DNA repair pathways. Although its action is essential to maintain genome integrity and to
protect against damage-induced mutations, as part of the NER and ICL machinery it can counteract the
effect of DNA damaging therapies such as platinum-based chemotherapy and radiotherapy. A promising
approach to enhance the effect of such therapies is to inhibit the action of DNA repair in cancer cells
using small molecules. In this work, we used a computational drug design workflow to provide a rational
design for novel compound 1 analogues, a lead inhibitor targeting the dimerization between XPF and
19
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 38
1
2
3
4
5
6
7
8
9
ERCC1, which is required for endonuclease activity. We identified seven compounds for which in-silico
simulations predicted attractive properties, namely binding affinities and ligand efficiency towards an
XPF site on the dimerization interface. The synthesis of the computationally designed compounds was
successfully carried out using a simple, robust and reproducible synthetic strategy. Interestingly,
compound 4 has good physicochemical properties, i.e. reasonable logP and ligand efficiency values, and
small molecular weight (see Table S4), and thus is a potential lead candidate for further optimization.
Following structure-activity relationship studies and in-vitro screening, this approach yielded compounds
3 and 4 as potent inhibitors of ERCC1-XPF activity. An in-vitro ERCC1-XPF endonuclease assay
identified compound 4 as the best ERCC1-XPF inhibitor with an IC50 value of 0.33 µM compared to
1.86 µM for 1. In addition, the Kd value for this compound was experimentally measured as 100 nM and
interacts with ERCC1-XPF in the subunit interaction domain and alters the related properties of this
heterodimerization interface. Compound 4 also showed a significant inhibition of the removal of
cyclobutane pyrimidine dimers compared with control cells after exposure of HCT-116 cells to UV
radiation and sensitized the cells to UV and cyclophosphamide-induced cytotoxicity, indicating
inhibition of NER and ICL repair. Furthermore, compound 4 possesses favourable pharmacokinetic
properties, particularly a lower LogD than compound 1, which means that it has a lower lipophilicity,
and greater metabolic stability. Our computational workflow successfully identified superior compounds
from a set of analogues differing by one substituent group. Additionally, it provided detailed information
about the ligand-protein interaction. Compound 4 was our best in vitro analogue. Detailed analysis of the
MM/GBSA binding energies, together with the visual analysis of the simulation results, suggests a
binding mode for compound 4 where the conserved hydrophobic core of the analogue is buried inside
the XPF binding site and, differently to the other active compounds (1 and 3) the positive R-substituent
is favorably exposed to the solvent. Although all of the compounds share the acridine moiety, our assay
showed that the inhibition activity of compound 4 was the consequence of the interaction of the drug
with the enzyme, and not the result of binding with DNA. The fact that compounds 4 and 5 (negative
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
20
ACS Paragon Plus Environment

Page 21 of 38
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
control) show drastically different activity despite possessing the same acridine moiety also provides
strong evidence that inhibition is not mediated via DNA intercalation. In future work, compound 4 will
be the starting point for further optimization towards a drug candidate compound to yield an optimized
lead analogue of compound 4. Since polar side-chain installation at the exterior part of the piperazine
position clearly enhances activity, we intend to explore a wider range of polar groups on the remote end
of the piperazine. At the same time, we will retain the dimethylaminoethyl group and explore structural
changes in parts of the molecule that are more deeply intercalated into the protein binding pocket.
Different generation of compound 4 derivatives will be synthesized according to the corresponding in
silico studies and the current SAR results in this study for the aim of obtaining a drug like candidate that
has superior physicochemical properties than compound 4. In summary, the use of in-silico methods to
design a superior compound to 1 led to compound 4 that can potentially be used in combination with
other existing DNA-damaging therapies to amplify their effects by sensitizing cancer cells.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Methods and Experimental section
General experimental procedures for preparation of the top compound 1 analogues
Synthesis and characterization of compounds 1-8 are described in Supporting Information, section 6.
Reactions were carried out in flame-dried glassware under a positive argon atmosphere unless otherwise
stated. Transfer of anhydrous solvents and reagents was accomplished with oven-dried syringes or
cannulae. Solvents were distilled before use. Chemicals were purchased from Sigma Aldrich Inc., and
were used without further purification. Thin layer chromatography was performed on glass plates pre-
coated with 0.25 mm silica gel. Flash chromatography columns were packed with 230-400 mesh silica
gel. Proton nuclear magnetic resonance spectra (1H NMR) were recorded at 500 MHz, and coupling
constants (J) are reported in hertz (Hz). Standard notation was used to describe the multiplicity of signals
1
observed in H NMR spectra: broad (br), multiplet (m), singlet (s), doublet (d), triplet (t), etc. Carbon
21
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 38
1
2
3
4
5
6
7
8
nuclear magnetic resonance spectra (13C NMR) were recorded at 125 MHz and are reported (ppm)
relative to the center line of the triplet from chloroform-d (77.0 ppm) or the center line of the heptuplet
from methanol-d4 (49.0 ppm). Infrared (IR) spectra were measured with a FT-IR 3000
spectrophotometer. Mass spectra were determined on a high-resolution electrospray positive ion mode
spectrometer. The high-performance liquid chromatography (HPLC) analyses were performed using a
Agilent 1100 LC/MSD instrument. Elution was done with a gradient of 10−95% solvent B in solvent A
(solvent A was 0.1% TFA in water, and solvent B was 0.1% acetic acid in MeCN) through an Agilent
column eclipse XDB- C18 (4.6× 250 mm, 5 μm) column at 1.0 mL/min. Area % purity was measured at
210 and 254 nm. The purity of the most active compound was assessed by HPLC (>95%).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Synthesis and characterization of inhibitors (1-8)
4-((6-Chloro-2-methoxyacridin-9-yl)amino)-2-((4-methylpiperazin-1-yl)methyl)phenol (1)
Synthesis of compound 1 was achieved through one pot sequential addition reaction in 3 steps. A mixture
of N-methyl piperazine (0.12 mL, 1.0 mmol), acetaminophen (0.15 g, 1.0 mmol) and 37% formaldehyde
(0.10 mL, 5 mmol) in isopropyl alcohol (2 ml) was taken in 10 mL single neck round bottom flask. Then,
the reaction mixture was stirred and heated under reflux at 65 °C for 6 h. Upon completion of the reaction
(monitored by TLC with 15% MeOH/DCM eluent system), the solvent was removed on a rotatory
evaporator. Next, the residue was dissolved in 3 mL ethanol and 3 drops of 12 M HCl were added.
Subsequently, the reaction mixture was heated at 90 °C under reflux for 90 min. Afterwards, 6,9-
dichloro-methoxyacridine (0.28 g, 1.0 mmol) was added to the mixture and further stirred at 90 °C under
reflux and the course of reaction followed by TLC until little or no starting material was detected (around
12 h). On cooling to room temperature, the reaction mixture was diluted with cold water and neutralized
to pH of 8-9 with 28% v/v ammonia solution. The alkaline solution was extracted with dichloromethane.
The organic layer was washed with brine, concentrated in vacuum, and purified by column
chromatography (gradient elution with 5% to 10% MeOH:DCM system) to afford compound 1 as orange
reddish semisolid (0.08 g) in 78 % yield; Rf 0.50 (2:8, MeOH:DCM); IR (cast film) νmax = 3272, 2923,
1629, 1254, 1231, 1032, 926, 816, 815, 775 cm-1; 1H NMR (500 MHz, CDCl3) δ 7.99 (s, 1H), 7.91 (d, J
= 8.7 Hz, 1H), 7.80 (d, J = 9.2 Hz, 1H), 7.29 (dd, J = 9.4, 2.6 Hz, 1H), 7.11 (d, J = 10.7 Hz, 1H), 7.03 (s,
1H), 6.77 – 6.70 (m, 2H), 6.49 (d, J = 2.5 Hz, 1H), 3.63 (s, 3H), 3.53 (s, 2H), 2.75 – 2.32 (m, 8H), 2.27
22
ACS Paragon Plus Environment

Page 23 of 38
1
Journal of Medicinal Chemistry
13
2
(s, 3H), OH and NH protons were not observed; C NMR (125 MHz, CDCl3) δ 155.9, 153.2, 147.8,
3
143.9, 137.0, 135.0, 125.0, 124.9, 122.0, 121.8, 120.0, 119.9, 119.7, 117.61, 116.8, 116.5, 116.1, 115.9,
100.3, 61.1, 55.2, 54.8 (2C), 52.4 (2C), 45.8; HRMS (ESI) calcd for C26H28ClN4O2 [M + H]+ 463.1895;
found 463.1890.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4-((6-Chloro-2-methoxyacridin-9-yl)amino)-2-((4-(3-methylbenzyl)piperazin-1-yl)methyl)phenol (2)
The previous method was employed to synthesize compound 2 with the following stoichiometric
amounts; 1-(3-methylbenzyl)piperazine (0.19 g, 1.0 mmol) and 6,9-dichloroacridine (0.28 g, 1.0 mmol)
to afford it as orange reddish semisolid (0.14 g) in 73 % yield; Rf 0.44 (1:9, MeOH:DCM); IR (cast film)
νmax = 3258, 2921, 1629, 1560, 1493, 1253, 1133, 1007, 927, 826, 775 cm-1; 1H NMR (500 MHz, CDCl3)
δ 8.05 (s, 1H), 7.95 (d, J = 9.2 Hz, 1H), 7.83 (d, J = 9.2 Hz, 1H), 7.32 (d, J = 7.6 Hz, 1H), 7.21 (t, J =
7.5 Hz, 1H), 7.17 – 7.06 (m, 5H), 6.87 (d, J = 7.1 Hz, 1H), 6.77 (d, J = 8.6 Hz, 1H), 6.62 (d, J = 1.9 Hz,
1H), 3.70 (s, 3H), 3.59 (s, 2H), 3.49 (s, 2H), 2.79 – 2.40 (m, 8H), 2.35 (s, 3H). OH and NH protons were
13
not observed; C NMR (125 MHz, CDCl3) δ 156.1, 153.9, 150.4, 146.8, 144.5, 137.9, 137.6, 136.3,
135.5, 129.9, 128.2, 127.9, 126.2, 125.4, 125.0, 122.1, 121.8, 120.6, 120.3, 119.2, 116.9, 116.5, 116.1,
115.9, 100.4, 62.8, 61.2, 55.4, 52.8 (2C), 52.5 (2C), 21.4; HRMS (ESI) calcd for C33H34ClN4O2 [M +
H]+ 553.2357; found 553.2365.
4-((6-Chloro-2-methoxyacridin-9-yl)amino)-2-((4-cyclohexylpiperazin-1-yl)methyl)phenol (3)
The previous method was employed to synthesize compound 3 with the following stoichiometric
amounts; 1-cyclohexyl piperazine (0.17 g, 1.0 mmol) and 6,9- dichloroacridine (0.28 g, 1.0 mmol) to
afford it as orange reddish semisolid (0.13 g) in 81 % yield; Rf 0.52 (0.5:9.5, MeOH: EtOAc); IR (cast
film) νmax = 3313, 2918, 1736, 1560, 1468, 1255, 1235, 1181, 1032, 929, 829, 722 cm-1; 1H NMR (500
MHz, CDCl3) δ 8.03 (d, J = 1.4 Hz, 1H), 7.92 (d, J = 9.6 Hz, 1H), 7.80 (d, J = 9.5 Hz, 1H), 7.28 (d, J =
5.7 Hz, 1H), 7.11 (d, J = 9.1 Hz, 2H), 6.91 (d, J = 9.3 Hz, 1H), 6.77 (d, J = 8.5 Hz, 1H), 6.69 (s, 1H),
3.71 (s, 3H), 3.60 (s, 2H), 2.73 – 2.22 (m, 8H), 1.83 (m, 4H), 1.63 (d, J = 12.8 Hz, 1H), 1.26 – 1.18 (m,
6H). OH and NH protons were not observed; 13C NMR (125 MHz, CDCl3) δ 156.1, 153.8, 146.9, 144.5,
136.3, 135.5, 125.4 – 116.8 (10 C), 101.3, 100.4, 98.8, 63.4, 61.3, 52.9 (2 C), 48.7 (2 C), 29.7, 28.9 (2
C), 26.2, 25.8 (2 C); HRMS (ESI) calcd for C31H35ClN4O2 [M + H]+ 531.2521; found 531.2527.
4-((6-Chloro-2-methoxyacridin-9-yl)amino)-2-((4-(2-(dimethylamino)ethyl) piperazin-1-yl) methyl)
phenol (4)
23
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 38
1
2
3
4
5
6
7
8
9
The previous method was employed to synthesize compound 4 with the following stoichiometric
amounts; 1-[2-(dimethylamino)ethyl] piperazine (0.16 g, 1.0 mmol) and 6,9-dichloroacridine (0.28 g, 1.0
mmol) to afford it as orange reddish semisolid (0.13 g) in 80 % yield; Rf 0.40 (1.3: 8.5: 0.2, MeOH:
DCM: Et3N); IR (cast film) νmax = 3361, 2918, 1736, 1562, 1467, 1295, 1255, 1143, 1032, 945, 828, 763,
722 cm-1; 1H NMR (500 MHz, CDCl3) δ 8.10 (s, 1H), 8.00 (s, 1H), 7.89 (d, J = 9.1 Hz, 1H), 7.40 (d, J =
9.8 Hz, 1H), 7.24 (d, J = 8.0 Hz, 1H), 7.09 (s, 1H), 6.85 (dd, J = 8.6, 2.7 Hz, 1H), 6.78 (d, J = 8.6 Hz,
1H), 6.56 (d, J = 2.3 Hz, 1H), 3.74 (s, 3H), 3.59 (s, 2H), 2.70 – 2.40 (m, 12H), 2.27 (s, 6H). OH and NH
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
13
protons were not observed; C NMR (125 MHz, CDCl3) δ 156.2, 153.6, 147.5, 144.0, 136.6, 135.2,
125.3, 125.1, 124.8, 122.0, 120.3, 120.2, 120.0, 119.8, 119.6, 117.5, 116.9, 116.3, 100.2, 61.2, 56.7, 56.7,
55.3, 53.3 (2C), 52.4 (2C), 45.8 (2C); HRMS (ESI) calcd for C29H35ClN5O2 [M + H]+ 520.2474; found
520.2473.
4-((6-Chloro-2-methoxyacridin-9-yl)amino)-2-((4-(3,4-dichlorophenyl)piperazin-1-yl)methyl)phenol
(5)
The previous method was employed to synthesize compound 5 with the following stoichiometric
amounts; 1-(3,4-dichlorophenyl) piperazine (0.23 g, 1.0 mmol), acetaminophen (0.15 g, 1.0 mmol), 37%
formaldehyde (0.10 mL, 5 mmol) and 6,9- dichloroacridine (0.28 g, 1.0 mmol) to afford it as orange
reddish semisolid (0.16 g) in 71 % yield; Rf 0.56 (0.5:9.5, MeOH:DCM); IR (cast film) νmax = 3271,
2918, 1736, 1563, 1227, 1180, 1080, 1031, 803, 721 cm-1; 1H NMR (500 MHz, CDCl3) δ 8.09 (s, 1H),
8.00 (s, 1H), 7.88 (d, J = 9.2 Hz, 1H), 7.40 (d, J = 9.3 Hz, 1H), 7.28 (d, J = 8.7 Hz, 1H), 7.24 (d, J = 9.0
Hz, 1H), 7.09 (s, 1H), 6.95 (d, J = 2.8 Hz, 1H), 6.87 (dd, J = 8.6, 2.7 Hz, 1H), 6.80 (d, J = 8.6 Hz, 1H),
6.73 (dd, J = 8.9, 2.8 Hz, 1H), 6.54 (d, J = 2.6 Hz, 1H), 3.74 (s, 3H), 3.63 (s, 2H), 3.27 – 3.15 (m, 4H),
13
2.72 – 2.62 (m, 4H), OH and NH protons were not observed; C NMR (125 MHz, CDCl3) δ 156.3,
153.1, 150.2, 147.9, 143.3, 137.0, 135.0, 132.9, 130.5, 125.2, 124.7, 122.9, 121.6, 121.4, 120.2, 120.0,
119.7, 117.9, 117.6, 117.0, 116.7, 116.2, 115.6, 115.5, 100.0, 61.2, 55.4 (2C), 52.2 (2C), 48.7; HRMS
(ESI) calcd for C31H28Cl3N4O2 [M + H]+ 593.1272; found 593.1272.
2-((4-(Bis(4-fluorophenyl)methyl)piperazin-1-yl)methyl)-4-((6-chloro-2-methoxyacridin-9-
yl)amino)phenol (6)
The previous method was employed to synthesize compound 6 with the following stoichiometric
amounts; 1-bis (4-fluorophenyl)methylpiperazine (0.29 g, 1.0 mmol), acetaminophen (0.15 g, 1.0 mmol),
37% formaldehyde (0.10 mL, 5 mmol) and 6,9- dichloroacridine (0.28 g, 1.0 mmol) to afford it as orange
24
ACS Paragon Plus Environment

Page 25 of 38
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
8
9
reddish semisolid (0.19 g) in 65 % yield; Rf 0.56 (0.5:9.5, MeOH:DCM); IR (cast film) νmax = 3275,
2919, 1737, 1566, 1255, 1217, 1181, 1032, 828, 722 cm-1; 1H NMR (500 MHz, CDCl3) δ 8.07 (s, 1H),
7.98 (d, J = 9.0 Hz, 1H), 7.84 (d, J = 9.2 Hz, 1H), 7.34 (m, 6H), 7.18 (d, J = 8.0 Hz, 1H), 7.06 (d, J = 1.8
Hz, 1H), 6.97 (m, 5H), 6.83 (dd, J = 8.5, 2.6 Hz, 1H), 6.75 (d, J = 8.6 Hz, 1H), 6.53 (d, J = 2.2 Hz, 1H),
4.23 (s, 1H), 3.69 (s, 3H), 3.57 (s, 2H), 2.74 – 2.21 (m, 8H); 13C NMR (125 MHz, CDCl3) δ 161.9 (2C),
156.2, 153.5, 150.3, 147.7, 143.7, 137.8 (2C), 136.7, 135.0, 129.2 (4C), 125.2, 124.8, 121.9, 121.7, 120.2,
119.9, 119.7, 117.6, 116.8, 116.5, 116.1, 115.5 (4C), 100.0, 74.3, 61.1, 55.3, 52.6 (2C), 51.5 (2C); HRMS
(ESI) calcd for C38H34ClF2N4O2 [M + H]+ 651.2333; found 651.2328.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4-((6-Chloro-2-methoxyacridin-9-yl)amino)-2-((4-(4-hydroxyphenyl)piperazin-1-yl)methyl)phenol
(7)
The previous method was employed to synthesize compound 7 with the following stoichiometric
amounts; 1-(4-hydroxyphenyl)piperazine (0.30 g, 1.0 mmol), acetaminophen (0.15 g, 1.0 mmol), 37%
formaldehyde (0.10 mL, 5 mmol) and 6,9- dichloroacridine (0.28 g, 1.0 mmol) to afford it as orange
reddish semisolid (0.17 g) in 59 % yield; Rf 0.58 (1.5:8.5, MeOH:DCM); IR (cast film) νmax = 3334,
2918, 1737, 1564, 1468, 1235, 1181, 1031, 930, 830, 722 cm-1; 1H NMR (500 MHz, CD3OD) δ 8.07 (d,
J = 9.3 Hz, 1H), 7.87 (d, J = 1.8 Hz, 1H), 7.82 (d, J = 9.3 Hz, 1H), 7.51– 7.46 (m, 1H), 7.41 (d, J = 2.5
Hz, 1H), 7.29 – 7.24 (m, 1H), 7.04 (dd, J = 8.7, 2.4 Hz, 1H), 6.93 – 6.89 (m, 1H), 6.88 – 6.81 (m, 3H),
6.70 (dd, J = 8.9, 2.3 Hz, 2H), 3.72 (s, 3H), 3.61 (s, 2H), 3.11 – 2.99 (m, 4H), 2.82 – 2.67 (m, 4H), OH
13
and NH protons were not observed; C NMR (125 MHz, CDCl3) δ 156.3, 153.5, 152.1, 150.2, 150.1,
145.3, 145.2, 144.0, 125.3, 124.7, 122.5, 121.9, 121.5, 120.4, 120.0, 118.8 (2C), 118 (2C), 118.4, 118.1,
117.0, 116.5, 116.1, 115.9, 61.3, 55.4, 52.7 (2C), 50.7 (2C); HRMS (ESI) calcd for C31H30ClN4O3 [M +
H]+ 541.2001; found 541.2010.
4-((6-Chloro-2-methoxyacridin-9-yl)amino)phenol (8)
21
Synthesis of compound 8 was carried out according to the reported procedures
with minor
modifications. The synthesis was performed by mixing 4-aminophenol (0.11 g, 1.0 mmol) and 6,9-
dichloroacridine (0.28 g, 1.0 mmol) in 10 mL single neck round bottom flask and stirred at 90 °C under
reflux. The course of reaction followed by TLC until little or no starting material was detected (around
12 h). On cooling to room temperature, the reaction mixture was diluted with cold water and neutralized
to pH of 8-9 with 28% v/v ammonia solution. The alkaline solution was extracted with dichloromethane.
The organic layer was washed with brine, concentrated in vacuum, and purified by column
25
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 38
1
2
3
4
5
6
7
8
9
chromatography (gradient elution with 1% to 5% MeOH:DCM system) to afford it as orange reddish
semisolid in 73 % yield; Rf 0.47 (1.5:8.5, MeOH:DCM); IR (cast film) νmax = 3325, 2918, 1737, 1562,
1468, 1236, 1180, 1103, 1050, 722 cm-1; 1H NMR (500 MHz, CD3OD) δ 8.12 (d, J = 9.4 Hz, 1H), 7.84
– 7.83 (m, 1H), 7.78 (d, J = 9.3 Hz, 1H), 7.61 (dd, J = 9.3, 2.6 Hz, 1H), 7.48 (d, J = 1.5 Hz, 1H), 7.35
(dd, J = 9.4, 2.1 Hz, 1H), 7.27 – 7.22 (m, 2H), 6.95 – 6.91 (m, 2H), 3.88 (s, 1H), 3.69 (s, 3H). OH proton
was not observed; 13C NMR (125 MHz, CD3OD) δ 158.9, 157.7, 141.7, 137.9, 133.2, 129.8, 128.4, 127.8
(2C), 126.6, 125.7, 122.2, 120.1, 119.3, 117.8 (2C), 116.1, 113.3, 104.5, 56.1.; HRMS (ESI) calcd for
C20H16ClN2O2 [M + H]+ 351.0908; found 351.0896.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Molecular docking of compound 1
25
The first step of our VS study was to investigate the binding mode of the lead compound 1 to the
pocket on the XPF C-terminus. We used twenty XPF structures as single targets for a relaxed complex
scheme (RCS) docking protocol, in order to accurately account for protein flexibility 34, 35. These
structures were extracted from the Protein Data Bank (PDB)36 entry 1Z00 17, reporting the NMR
ensemble of the dimerization complex between the ERCC1 and XPF HhH2 domains. The XPF structures
were optimized with a minimization process, details of which are reported in Supporting Information,
section 1.1. The binding site of each target was defined as the geometric center of the residues Y833,
N834, P837, Q838, M856, K860, N861 and I862 on the XPF HhH2 domain, as reported in our previous
study 25, where the residues are numbered according to the 1Z00 PDB file. Molecular docking
simulations were performed using the Lamarckian Genetic Algorithm (LGA) 37 and the built-in scoring
38
function
of Autodock4 39. Technical details regarding the docking simulations are reported in
Supporting Information, section 1.2.
Characterization of the binding pocket and pharmacophore modeling
To characterize the binding site and the binding mode of 1, electrostatic maps were generated in the
40
MOE2015 software package , based on the docked pose of 1, using -2 kcal/mol for the potential
isosurfaces of hydrophobic, acceptor and donor probe atoms, and the Poisson-Boltzmann equation to
26
ACS Paragon Plus Environment

Page 27 of 38
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
compute the potentials. The maps were built within 4.5 Å of the ligand pose obtained from docking.
Using the information from the docking and electrostatic mapping, a pharmacophore model for 1 was
built with the MOE Pharmacophore Query Editor and the EHT scheme 41.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Docking-based virtual screening of compound 1 analogues
Fifty-seven analogues of compound 1 structure were designed by replacement of the piperazine N-methyl
group with other moieties expected to capture additional binding interactions of the piperazine ring. The
42
structures were prepared using Schrödinger LigPrep to account for multiple tautomers, protonation
states and low-energy ring conformations, using the same approach as for 1 (Supporting Information,
section 1.2). The molecular docking simulations were performed with MOE Dock. Only the XPF
structure from the best compound 1 complex was used as a target for the VS study. Different
conformations of the analogues were obtained using Conformation Import. The previously built
pharmacophore model was used for the placement step, in which 30 poses are returned according to the
London dG scoring method 43. To account for the local arrangement of the pocket residues upon ligand
binding, the Induced Fit method was selected for the refinement step, where the side chains of the binding
pocket were left to move freely. At the end, one pose scored with the Generalized Born Volume
Integral/Weighted Surface Area (GBVI/WSA) function was returned 43. Water/octanol partition
44
coefficients (logP) of the molecules were calculated in MOE using the SlogP function , which takes
into account the given protonation state of the molecule under examination. Ligand efficiencies were
calculated as the ratio between the computed binding energies and the number of heavy atoms of each
analog.
Molecular dynamics simulations and MM/GBSA rescoring of the analogues
To calculate an average binding energy for the hits, we performed 2 ns of MD simulations for the top
ranked compound-XPF complexes. Details of the preparation method and the parameters used in these
27
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 38
1
2
3
4
5
6
7
8
9
simulations are reported in Supporting Information, section 2.1. Free energy calculations were performed
over the trajectories with the MM/GBSA method, using the MMPBSA.py script 45. For technical details
regarding the calculations refer to Supporting Information, section 2.1. The calculations were performed
on snapshots extracted every 10 ps from the MD trajectories, and per-residue decompositions of the
binding energies were also performed for the residues within 10 Å of any analog atom at the beginning
of the simulations. We also calculated the entropic contribution of ligand binding using the normal mode
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
46
analysis (NMA) method . The final binding energies used to rank the compounds were calculated
combining the MM/GBSA and entropy contributions with the equation S1 in Supporting Information.
ERCC1-XPF protein preparation
Human ERCC1-XPF wild-type protein was obtained as previously described 29. Basically, the
recombinant protein was expressed from a bicistronic plasmid (kindly provided by Dr. Richard Wood,
University of Texas MD Anderson Cancer Center, Smithville, TX) in the E. coli BL21 (DE3) strain.
Since both XPF and ERCC1 contained a polyhistidine (His-6) tag, the proteins extracted from E. coli
were incubated with a ProBond Nickel-Chelating Resin (Thermo Fisher Scientific). Protein eluted from
the Ni affinity column was subsequently loaded into a Hi-trap heparin column (GE Healthcare). Fractions
recovered from the heparin column that contained ERCC1-XPF were dialyzed, concentrated and stored
at -80ºC in 10 mM HEPES pH 7.4, 2.5 mM β-mercaptoethanol, 0.01% CHAPS, 0.25 mM EDTA, 50%
glycerol and 25 mM NaCl.
Microplate fluorescence incision assay
We followed a previously described protocol 2, 29. Briefly, reactions were carried out in 384-well black,
flat-bottomed microtiter plates (OptiPlate – 384 F, Perkin Elmer) in a total volume of 20 µl containing
the indicated concentrations of inhibitor compounds, 100 nM stem-loop substrate [6-FAM-5′-
CAGCGCTCGG(20T)CCGAGCGCTG-3′-dabcyl], 25 ng ERCC1-XPF, 50 mM Tris-Cl pH 8, 20 mM
28
ACS Paragon Plus Environment

Page 29 of 38
1
Journal of Medicinal Chemistry
2
3
4
5
6
7
NaCl, 0.5 mM DDT and 0.75 mM MnCl2 at 25ºC. Fluorescent readings were obtained using a FLUOstar
Optima fluorimeter (BMG Labtech) with Optima software at an excitation of and emission wavelengths
of 485 and 520 nm respectively, for 12 minutes.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Steady-state fluorescence assays
Steady-state fluorescence spectra were measured at room temperature on a Perkin-Elmer LS-55
spectrofluorometer (Freemont, CA) with 5 nm spectral resolution for excitation and emission using 30-
80 nM solution of purified recombinant ERCC1-XPF protein complex. Protein fluorescence was excited
at 295 nm, and fluorescence emission spectra were recorded in the 300-400 nm range: changes in
fluorescence intensity was monitored at the emission maximum (330 nm). In studying the effects of
inhibitors on protein fluorescence intensities, additions to protein samples were made from inhibitors
stock solutions, keeping the protein dilution below 3%.
Expression and purification of recombinant truncated forms of ERCC1 and XPF:
Synthetic DNA sequences encoding for human ERCC1 codons 96-297 (dERCC1) and XPF codons 667-
916 (dXPF) inserted into separate pET-28a expression plasmids were used to transform BL21(DE3) cells
2
and express the corresponding recombinant proteins . Since the truncated form of the proteins both
contained N-terminal polyhistidine-tags (His-tags) they were purified utilizing a nickel-chelating resin
(ThermoFisher Scientific) under conditions similar to the ones described 2. Both truncated peptides were
finally solubilized to final concentrations for the experiments of 40 nM dXPF and 20 nM dERCC1 in a
buffer (MST buffer) containing: 20 mM HEPES pH 7.4, 50 mM NaCl, 1 mM DTT, 2 mM MgCl2, 5%
glycerol and 0.01% CHAPS.
Microscale thermophoresis (MST) measurements
We tried to remove the His-tag from one of the truncated peptides (preferentially dERCC1 because of its
low yield) to carry out MST equilibrium experiments. However, the peptides proved to be resistant to
29
ACS Paragon Plus Environment