Design, synthesis and in vitro cell-free/cell-based biological evaluations of novel ERCC1-XPF inhibitors targeting DNA repair pathway

Article abstract of DOI:10.1016/j.ejmech.2020.112658

The structure-specific ERCC1-XPF endonuclease is essential for repairing bulky DNA lesions and helix distortions induced by UV radiation, which forms cyclobutane pyrimidine dimers (CPDs), or chemicals that crosslink DNA strands such as cyclophosphamide and platinum-based chemotherapeutic agents. Inhibition of the ERCC1-XPF endonuclease activity has been shown to sensitize cancer cells to these chemotherapeutic agents. In this study, we have conducted a structure activity relationship analysis based around the previously identified hit compound, 4-((6-chloro-2-methoxyacridin-9-yl)amino)-2-((4-methylpiperazin1-yl)methyl)phenol (F06), as a reference compound. Three different series of compounds have been rationally designed and successfully synthesized through various modifications on three different sites of F06 based on the corresponding suggestions of the previous pharmacophore model. The in vitro screening results revealed that 2-chloro-9-((3-((4-(2-(dimethylamino)ethyl)piperazin-1-yl)methyl)-4-hydroxyphenyl)amino)acridin-2-ol (B9) has a potent inhibitory effect on the ERCC1-XPF activity (IC₅₀ = 0.49 μM), showing 3-fold improvement in inhibition activity compared to F06. In addition, B9 not only displayed better binding affinity to the ERCC1-XPF complex but also had the capacity to potentiate the cytotoxicity effect of UV radiation and inhibiting the nucleotide excision repair, by the inhibition of removal of CPDs, and cyclophosphamide toxicity to colorectal cancer cells.

Full text of DOI:10.1016/j.ejmech.2020.112658

Journal Pre-proof
Design, synthesis and in vitro cell-free/cell-based biological evaluations of novel
ERCC1-XPF inhibitors targeting DNA repair pathway
Ahmed H. Elmenoufy, Francesco Gentile, David Jay, Feridoun Karimi-Busheri,
Xiaoyan Yang, Olivier M. Soueidan, Rajam S. Mani, Gloria Ciniero, Jack A.
Tuszynski, Michael Weinfeld, Frederick G. West
PII:
S0223-5234(20)30630-9
DOI:
https://doi.org/10.1016/j.ejmech.2020.112658
EJMECH 112658
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 8 June 2020
Revised Date: 7 July 2020
Accepted Date: 8 July 2020
Please cite this article as: A.H. Elmenoufy, F. Gentile, D. Jay, F. Karimi-Busheri, X. Yang, O.M.
Soueidan, R.S. Mani, G. Ciniero, J.A. Tuszynski, M. Weinfeld, F.G. West, Design, synthesis and in vitro
cell-free/cell-based biological evaluations of novel ERCC1-XPF inhibitors targeting DNA repair pathway,
European Journal of Medicinal Chemistry, https://doi.org/10.1016/j.ejmech.2020.112658.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. 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.
© 2020 Elsevier Masson SAS. All rights reserved.

Design, synthesis and in vitro cell-free/cell-based biological evaluations of
novel ERCC1-XPF inhibitors targeting DNA repair pathway
Ahmed H. Elmenoufy1,2, Francesco Gentile3¥, David Jay4, Feridoun Karimi-Busheri4, Xiaoyan
Yang4, Olivier M. Soueidan1, Rajam S. Mani4, Gloria Ciniero5,6, Jack A. Tuszynski3,4, Michael
Weinfeld4 and Frederick G. West1*
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
5Department of Mechanical and Aerospace Engineering, Politecnico di Torino, Torino, Italy
10129
6Università di Torino, Torino, Italy 10124
¥Current address: Vancouver Prostate Centre, University of British Columbia, Vancouver,
British Columbia, Canada V6H 3Z6
* Corresponding author.
E-mail address: fwest@ualberta.ca

Abstract
The structure-specific ERCC1-XPF endonuclease is essential for repairing bulky DNA lesions
and helix distortions induced by UV radiation, which forms cyclobutane pyrimidine dimers
(CPDs), or chemicals that crosslink DNA strands such as cyclophosphamide and platinum-based
chemotherapeutic agents. Inhibition of the ERCC1-XPF endonuclease activity has been shown to
sensitize cancer cells to these chemotherapeutic agents. In this study, we have conducted a
structure activity relationship analysis based around the previously identified hit compound, 4-
((6-chloro-2-methoxyacridin-9-yl)amino)-2-((4-methylpiperazin1-yl)methyl)phenol (F06), as a
reference compound. Three different series of compounds have been rationally designed and
successfully synthesized through various modifications on three different sites of F06 based on
the corresponding suggestions of the previous pharmacophore model. The in vitro screening
results revealed that 2-chloro-9-((3-((4-(2-(dimethylamino)ethyl)piperazin-1-yl)methyl)-4-
hydroxyphenyl)amino)acridin-2-ol (B9) has a potent inhibitory effect on the ERCC1-XPF
activity (IC50 = 0.49 µM), showing 3-fold improvement in inhibition activity compared to F06.
In addition, B9 not only displayed better binding affinity to the ERCC1-XPF complex but also
had the capacity to potentiate the cytotoxicity effect of UV radiation and inhibiting the
nucleotide excision repair, by the inhibition of removal of CPDs, and cyclophosphamide toxicity
to colorectal cancer cells.
Keywords:
ERCC1-XPF small molecule inhibitors, Targeted DNA repair therapy, Colorectal cancer cells
sensitisation, Cyclophosphamide potentiation, Molecular docking.

1. Introduction
ERCC1-XPF is a heterodimeric, highly conserved, structure-specific endonuclease that serves a
central function in several critical DNA repair pathways. These include nucleotide excision
repair, interstrand crosslink (ICL) repair, and double-strand break (DSB) repair 1-3. Accordingly,
ERCC1-XPF occupies a pivotal role in the response of tumors to DNA-damaging
chemotherapeutic drugs such as platinum-containing compounds, cyclophosphamide and
mitomycin C (MMC) 4, 5. The enzyme has also an important role of removing helix-distorting
DNA lesions and bulky adducts resulting from UV damage, including cyclobutane pyrimidine
6
dimers (CPDs) and pyrimidine-(6,4)-pyrimidone photoproducts (6-4PPs) . A recent study
revealed that 72% of patients with colorectal cancer exhibited high expression levels of ERCC1
7
protein , suggesting that inhibitors of ERCC1-XPF could play an important role in increasing
the sensitivity of cancer cells to chemotherapeutic agents targeting DNA, which would positively
impact the survival rate of cancer patients in the clinic. Thanks to the availability of X-ray crystal
structures of multiple sites in ERCC1-XPF that could be targeted by small molecules to inhibit
the endonuclease activity 8, the enzyme complex has become an objective for drug design studies
9-13
.
The ERCC1 monomer of ERCC1-XPF is catalytically inactive but functions as a regulator of
DNA–protein and protein–protein interactions. On the other hand, XPF contains the
endonuclease active site as well as an inactive helicase-like motif, and engages in additional
protein–protein interactions and DNA binding 14. However, enzymatic activity requires the
dimerization of the two proteins because neither protein alone is stable, and both undergo rapid
degradation in the cell when not part of the stable heterodimer 8, 15, 16. Dimerization between

ERCC1 and XPF is mediated by the interaction of double helix−hairpin−helix (HhH2) motifs
found in the C-terminal domains of each protein. Small molecule inhibitors able to disrupt the
interaction domains linking XPF and ERCC1 would be expected to block heterodimerization,
and hence reduce DNA repair activity. As a consequence, cancer cells would be sensitized to
conventional DNA crosslinking chemotherapeutic treatments and radiotherapy.
Several X-ray crystal structures of the HhH2 interaction domains (PDB code 2A1J and 1Z00)
permit rational drug design studies to develop small molecules to disrupt the heterodimerization
17
interface. Our initial efforts, which focused on the XPA active site
and the XPA−ERCC1
interaction site 18, validated the computer-aided drug design as an approach to identify and
optimize inhibitors for DNA repair processes involving NER 19; however, because the intact
ERCC1-XPF heterodimer is central to a diverse set of DNA repair pathways such as ICL, DSB
and NER, whereas XPA is involved mainly in NER, we regarded targeting of the dimerization
interface as more promising and general 14. Several studies have been carried out recently to
develop ERCC1-XPF inhibitors. Therapeutic peptides that mimic the ERCC1 HhH2 domain
(residues 220−297) were reported by Yang et al. as a potential strategy to inhibit the NER
pathway and potentiate DNA-damaging agents to cancer cells. 20 However, the translation of this
10
modality into the clinic is challenging in terms of the cost and applicability. Arora et al.
,
9
12, 13
Heyza et al. , and Chapman et al.
have described small molecules that target the XPF
endonuclease active site. Some of these endonuclease inhibitors, with IC50 values in the
nanomolar range, diminished NER activity and enhanced the cytotoxic effect of platinum-
containing drugs in cell culture. A possible drawback of the use of inhibitors operating through
metal ion chelation is the similarities among active sites of cellular endonucleases, which raises
14
the potential for off-target interactions. . McNeil et al. used in-silico screening to target three

different sites on the XPF HhH2 domain with the aim of identifying potential inhibitors that
could stably bind to it and hinder the interaction between ERCC1 and XPF domains, as well as
binding to the XPF endonuclease domain 21. They demonstrated that one of their compounds
bound to the XPF domain in vitro, inhibited the NER pathway in melanoma cells, and sensitized
such cells to cisplatin. However, the medium to high micromolar range for the IC50 and Kd
11
values of this active compound indicated the need for further optimization. Jordheim et al.
developed small molecule ERCC1-XPF inhibitors acting through the XPF interaction domain,
which led to heterodimerization disruption in vitro. They also reported that F06 (also called
NERI02 or NSC-130813 11, Fig. 1) sensitized cancer cells to cisplatin and MMC. This result
affirmed the notion that enhancing the cytotoxicity of cross-linking agents could be achieved via
targeting ERCC1-XPF protein-protein interaction. However, the toxicity and potency of F06
11
were deemed suboptimal, and Jordheim et al. suggested a derivatization strategy to optimize
the compound was required. In addition, most of the previous efforts to develop ERCC1-XPF
inhibitors were tested using a truncated form of the heterodimer, and not the full-length protein 9,
10, 12, 13, 18, 21. It has been shown that the truncated form of ERCC1-XPF is ~15-fold less active
than the full length ERCC1−XPF protein under standard reaction conditions 14, 15. Using F06 as a
reference compound, computer-aided drug design techniques were employed to identify
promising structural analogues. Electrostatic mapping of F06 binding pocket, pharmacophore
modeling, molecular docking, and molecular dynamics (MD)-based rescoring were used to rank
F06 analogues by their predicted binding affinity values for the XPF domain 22. One of the
derivatives, which displayed good physicochemical properties, had the ability to inhibit removal
of CPDs in cells irradiated with UV as well as increased the capacity to potentiate
22
cyclophosphamide and UV radiation towards colorectal cancer cells relative to the initial hit .

More recently, we conducted a study following a computational workflow to investigate one of
the potential binding sites of compound F06 and propose potential candidates to be synthesized
23
and tested in vitro . The results of the study yielded one compound that showed an IC50 value
of 0.49 µM for the inhibition of ERCC1-XPF endonuclease activity, a significant improvement
compared to the 1.86 µM IC50 value determine for F06. With a potential lead compound and a
robust computational approach, we sought to further develop inhibitors of DNA repair targeting
ERCC1-XPF. In this study, we report an extended structure activity relationship analysis through
synthesis of three different series of compounds based on modification of specific sites on the
structure of F06. The syntheses of all compounds were achieved through a straightforward and
convergent route, and their inhibitory activity against the full-length ERCC1-XPF was tested in
vitro. The most active compound was then assessed further for its inhibition of repair of UV-
induced CPDs in human colorectal cancer cells, as well as its activity as a sensitizing agent to
UV irradiation and cyclophosphamide. Notably, although it was developed in the 1950s,
cyclophosphamide continues to be used clinically in the treatment of lymphomas and other
cancers and is now being examined as part of a chemotherapeutic combination for colorectal
cancer in a metronomic setting 24, 25
.
2. Results and discussion
2.1 Chemistry
In our previous work, we reported the first generation (Gen A) of F06 (1) analogues based on
the modification of the piperazine ring with solvent-exposed moieties according to our reported
computational studies 22. In the present work, we focused on further modifying three different
sites of F06 (1) according to our suggested pharmacophore model and investigating their
structure activity relationship as shown in Fig. 1.

Figure 1. Structure of F06 (1) hit and the potential modification sites for hit to lead optimisation.
For the second generation series (Gen B) including controls (Fig. 2), we modified the acridine-
methoxy group of F06 by conversion to a polar hydrogen bond donor group such as OH (B2) or
by esterification (B6) or alkylation (B7 and B8) to gain more insight into the role of this group
on the activity of ERCC1-XPF.
The Gen B compounds were prepared via reaction of the phenolic OH of B5 with acyl chloride
or benzyl chloride derivatives to afford compounds B6, B7 and B8, respectively as depicted in
Scheme 1. B2 synthesis was prepared by dealkylation of B3 (2-methoxy-6,9-chloroacrdine) with
BBr3. Syntheses of F06 (1), B5 and B9 were carried out according to the reported one-pot
22
procedure with minor modifications, as shown in Scheme 1. Beginning with a Mannich-like
reaction of p-acetamidophenol with secondary amine and formaldehyde, the resulting viscous
residue after removal of unreacted formaldehyde was deacetylated at the anilino position with 6
M HCl. Finally, addition of 2-methoxy-6,9-chloroacridine or 2-hydroxy-6,9-chloroacridine
afforded compounds F06 (1) and B5, or B9, in good yields after isolation.

NH2
Cl
Cl
N
OH
O
N
Cl
N
Cl
N
OH
B2
B3
B4
N
N
N
N
N
N
HO
Cl
HO
HO
Cl
NH
N
NH
NH
O
OH
O
O
Cl
N
N
F06 (1)
B5
B6
N
N
N
N
N
N
N
HO
O
HO
NH
NH
NH
O
O
OH
Cl
N
Cl
N
Cl
N
B7
B8
B9
Figure 2. Structures of Gen B compounds including controls.

1) Multi-component reaction
O
O
NH2
OH
R
NH
OH
NH
6 M HCl
1 h, reflux, 70 °C
N
R
O
R
i-PrOH, 60 °C
N
N
Ethanol
85 %
N
H
H
N
12 h
N
H
71 %
OH
B4
R
N
2) SNAr
NH2
N
HO
Cl
R
OH
Ethanol, HCl
Reflux
N
NH
N
OH
N
100 °C
66 %
Cl
N
OH
B2
Cl
N
B9
B5
R=
R=
CH3
3) Nucleophilic substitution
Et3N/ DCM
Or
NaH/ DMF
O
Cl
Cl
Cl
B6
B5
B5
O
1 h
Cl
N
Et3N/ DCM
Or
NaH/ DMF
B7 and B8
B3
1 h
Scheme. 1. General synthetic route of Gen B compounds.
As a probe of the significance of aniline substitution pattern, especially the presence of the
piperazine moiety found in 1 and most of the Gen B compounds, the third generation (Gen C),
compounds were derived from condensation of the chloro-acridine building block with a variety
of readily available substituted anilines containing a variety of substituents, including moieties
such as heterocyclic, branched alkyl, aliphatic and aromatic groups (Fig. 3). Synthesis of Gen C
compounds proceeded through nucleophilic aromatic substitution of B2 and different substituted
aniline derivatives as shown in Scheme 2.

F
F
F
NH
NH
NH
OH
OH
OH
Cl
N
Cl
N
Cl
N
C1
C2
C3
O
N
NH
N
NH
N
OH
OH
Cl
Cl
C4
C5
HO
Cl
NO2
O
O
CN
NH
O
O
NH
NH
OH
OH
OH
N
Cl
N
Cl
N
D6
C7
C8
O
HS
Cl
NH
NH
OH
OH
Cl
N
N
C9
C10
Figure 3. Chemical structures of Gen C analogues.

R4
R1
R3
R2
R5
NH
R4
R1
Cl
EtOH, 90 °C
2-4 h
R3
R2
R5
OH
OH
HCl
96-98 %
NH2
Cl
N
Cl
N
B2
Gen C generic structure
Cl
O
Cl
N
B3
Scheme. 2. General synthetic route for Gen C compounds
Gen D compounds were designed to probe the importance of the tricyclic acridine unit by
replacing it with a quinoline (Fig. 4). The role of these functionalities was investigated on the
activity of ERCC1-XPF compared to the hit compound. D2 and D3 were used as controls. D2
was synthesized via nucleophilic aromatic substitution of D1 (4-chloro-7-methoxyquinoline) and
4-aminophenol. D1 and D3 were purchased from Sigma Aldrich Inc. in high purity. D4 was
synthesized following the experimental procedure for B9 except using D1 in place of the usual
acridine partner in the final SNAr coupling step.

N
N
N
OH
HO
O
OH
N
Cl
N
HN
N
NH
HN
N
2HCl
O
O
O
N
D1
D2
D3
D4
Figure 4. Chemical structures of Gen D analogues.
2.2 ERCC1−XPF Endonuclease Inhibition Assay.
22
As reported previously, an in vitro assay based on real-time change in fluorescence can be
used to evaluate the inhibitory effect of the derivatives on the endonuclease activity of
26
ERCC1−XPF. This assay uses a stem−loop substrate (consisting of a 20mer-oligo-dT single-
strand loop and a 10mer-duplex stem), labeled on the 3′-terminus with the quencher dabcyl and
on the 5′-terminus with the 6-FAM fluorophore. Upon cleavage of the 10mer-duplex region by
ERCC1−XPF, a product consisting of an 8-base 5′- labeled FAM fragment is released, which
results in a time-dependant increase in the fluorescent signal.
2.2.1 Inhibitory effect of Gen B compounds on ERCC1-XPF activity
As shown in Fig. 5, two compounds, B5 and B9 showed inhibited ERCC1−XPF nuclease
activity significantly more than F06. B5 differs structurally from F06 only via the presence of a
phenolic OH in place of the OMe found in F06, suggesting that a hydrogen bond donor at this
position is beneficial for binding. The same trend is seen in the comparison of the simple
acridines B2 and B3, although at significantly reduced activities relative to B5 and B9.
Fragments consisting of either only the acridine (B3) or only the piperazine-substituted

aminophenol (B4) resulted in complete loss of inhibition. To measure the half-maximum
inhibitory concentrations (IC50) for the best inhibitor B9 on the full-length protein, the initial
velocity (Vo, change in relative fluorescence units (RFU)/time) of the enzyme was plotted
against various concentrations of the drug (Fig. 5B). The IC50 of B9 was estimated as 0.49 ± 0.04
μM (consistent with our previously reported value for this compound 23), which was
approximately 4 times more efficient than F06, which has an estimated IC50 of 1.86 ± 0.25 μM
22. The improved inhibitory activity of B9 indicates the importance of replacement of the
acridine methoxy group with a phenolic hydrogen bond donor moiety, along with incorporating a
polar extension at the distal nitrogen atom of the piperazine ring, to overall inhibitory activity.
A

Figure 5. In vitro screening of inhibition of ERCC1−XPF endonuclease activity by Gen B
compounds. Measurements of endonuclease activity of ERCC1−XPF were carried out using a
microplate fluorescent reader. The increase in fluorescence (RFU) signal with time resulted from
the release of a fluorescent 8-base fragment upon the incubation of the DNA stem−loop substrate
with ERCC1−XPF. A) Time-dependent plot of fluorescence in the presence of each Gen B
compound (10 μM each). B) Plot of enzyme rate (ΔRFU/time) vs B9 concentration.
2.2.2 Inhibitory effect of Gen C compounds on ERCC1-XPF activity
The purpose for designing Gen C compounds was to study the aminophenol structure activity
relationship without incorporating the piperazine moiety to simplify the structures but retain the
acridine with the phenolic group. Afterwards, comparing their activity with the hit F06 and the
best inhibitor from the B series of compounds, B9. Different aniline substituents, with various
structural features common to bioactive small molecules 27, were selected for probing various
substitution patterns to screen if there are any identifiable trends,. Unfortunately, the inhibitory
effect of Gen C compounds on ERCC1-XPF activity was substantially less than either B9 or F06
(1) as depicted in Fig.6. However, C7 showed modest inhibition of the endonuclease activity.
These results indicate that the presence of the piperazine ring is critical for maintaining
inhibitory activity regardless of the substitution on the aminophenol ring.

Figure 6. Relative in vitro inhibition of the rate of ERCC1−XPF endonuclease activity
(ΔRFU/time) by Gen C compounds (10 μM each) vs control (no inhibitor).
2.2.3 Inhibitory effect of Gen D compounds on ERCC1-XPF activity
Our previous computational studies of F06 indicated that the planar aromatic rings of the
acridine moiety embedded in the XPF binding pocket can engage in favorable π-π interactions
with aromatic residues in the XPF HhH2 domain, and contribute to the binding energy and
ligand efficiency of F06 22, 23. However, this has not been investigated experimentally to
determine whether reduced hydrophobic surface area with less extended aromatic units would
impact the inhibition of ERCC1-XPF. To probe this question, we chose to replace the acridine
moiety with a similarly substituted quinoline. As it can be seen in Fig. 7, D1 and D2 failed to
inhibit ERCC1-XPF. Although D3 and D4 displayed modest inhibitory activity, it was less than
F06 and Gen B9, indicating the importance of retaining an extended tricyclic moiety to achieve

high inhibition. Interestingly, the most active member of this series, D4, bears the same
dimethylaminoethyl substituent on the piperazine ring as is found in B9, supporting the
significant role this moiety plays in ERCC1-XPF inhibition.
2500
D1
D2
2000
Control
1500
D3
D4
1000
F06
500
B9
0
0
2
4
6
minutes
Figure 7. In vitro screening of inhibition of ERCC1−XPF endonuclease activity by Gen D
compounds. (See Figure 5 legend for details.)
2.3 B9 binding to ERCC1−XPF
Since B9 displayed the most effective inhibition in vitro it was subjected to further analysis. The
binding affinity of B9 (active compound) and B7 (negative control) to ERCC1−XPF was
measured using intrinsic fluorescence spectroscopy by monitoring the protein tryptophan
22
fluorescence quenching as previously described . The observed fluorescence quenching at 330
nm upon the addition of 4 μM B7 was less than 2% (Fig. 8A), indicating no evidence of
significant small molecule-protein interaction. Conversely, treatment with B9 (4 μM) induced
nearly 25% ± 3% quenching of fluorescence (tryptophan residues) at 330 nm, evidence of a

robust interaction of B9 with the ERCC1−XPF heterodimer (Fig. 8B). The Kd value of B9 for
ERCC1−XPF complex was experimentally determined by plotting ligand concentration as a
function of fluorescence quenching. A representative plot of the concentrations of B9 versus the
relative fluorescence intensity is depicted in Fig. 8C (inset). Nonlinear regression analysis of
28
these data was performed as described in our earlier study, and the results showed unimodal
binding with a Kd value of 85 ± 5 nM (Fig. 8C and Table 1).
A
C
1.0
500
400
300
200
100
0
ERCC1.XPF
ERCC1.XPF + B7 (4µM)
250
200
150
100
Kd = 85 ± 5 nM
0.5
0
500 1000 1500 2000 2500 3000
[Compound B9], nM
0.0
1.0
300
325
350
375
400
0
500 1000 1500 2000 2500 3000
[Compound B9], nM
Wavelength, nm
B
D
600
400
200
0
ERCC1.XPF
ERCC1.XPF + B9 (4µM)
350
Kd = 1.05 ± 0.05 µM
300
0.5
0.0
250
200
150
0
2000
4000
6000
8000
[Compound B9], nM
300
325
350
375
400
0
2000
4000
6000
8000
Wavelength. nm
[Compound B9], nM
Figure 8. Binding affinity (Kd) measurements between ERCC1-XPF complex and B9 (active
inhibitor) and B7 (negative control). Quenching of intrinsic protein fluorescence (tryptophan
residues) was utilized to trace interaction between inhibitors and protein complex. (A)
Tryptophan fluorescence of ERCC1-XPF (50 nM) incubated with B7 (negative control). (B)
Tryptophan fluorescence of ERCC1-XPF (50 nM) incubated with B9 (active inhibitor). (C) Plot

of ERCC1-XPF fluorescence quenching vs concentration of B9 to determine unimodal binding
pattern and Kd (85 nM). Fluorescence intensity was monitored at 345 nm upon protein excitation
at 295 nm (see inset). (D) Plot of stem-loop substrate (lacking the quencher) fluorescence
quenching (100 nM) vs concentration of B9 to determine unimodal binding pattern and Kd (1.05
± 0.05 μM). Fluorescence intensity was monitored at 520 nm upon DNA substrate excitation at
490 nm (see inset).
Table 1
The half-maximum inhibitory concentrations (IC50) and binding constant values of inhibitors.
The IC50 values are the averages of data from at least three different experiments of Vo versus
compound concentration.
Compound
F06 (1)
B7
Kd protein (nM)
140 ± 5
Kd DNA (nM)
N.D.*
IC50 (µM)
1.86 ± 0.25
N.D.*
N.D.*
200 ± 10
85 ± 5
1050 ± 50
0.49 ± 0.04
B9
* Not determined.
All of the compounds in Gen B except for B4 contain an acridine unit, and this moiety is known
to bind to DNA through intercalation. The possibility that the observed activity of potent
inhibitors such as B9 could result from interactions with the stem-loop substrate had to be
considered. To determine if the binding to the DNA substrate plays a major role in the inhibitory
activity of B9, we compared the interaction of B9 and B7 with the stem-loop substrate (lacking
the dabcyl quencher) by following the change in fluorescence with increasing concentration of
inhibitor. We observed that B9 has a binding affinity for the DNA stem−loop (Kd = 1.05 ± 0.05

µM, Fig. 8D, Table 1), a value that is nearly 12-fold lower than what was observed for
ERCC1−XPF. This difference in binding affinity values offers evidence that inhibition of
endonuclease activity by B9 results from its interaction with ERCC1−XPF complex, not with the
DNA substrate. Notably, B7 showed a unimodal binding with a Kd value of 200 ± 10 nM (Fig.
S1). Thus, the negative control B7, which shares significant structural similarity with B9, binds
more tightly to the DNA substrate, but displays no inhibition of the nuclease activity of ERCC1-
XPF. This provides further support that DNA binding interactions by these small molecules is
not an important contributor to their inhibitory activity. We have also previously shown, using
microscale thermophoresis, that a compound with a structure similar to B9 interacted with
ERCC1−XPF in the subunit interaction domain 22. This earlier observation is consistent with the
hypothesis that the inhibition of ERCC1-XPF endonuclease activity by B9 is mediated by
disruption of heterodimerization between ERCC1 and XPF.
2.4 Inhibition of DNA repair by B9
To test for inhibition of nucleotide excision repair, we monitored the cellular removal of UVC-
induced cyclobutane pyrimidine dimers (CPDs) in HCT 116 cells over a 24-hour period using an
immunofluorescence assay. The results (Fig. 9) show that B9 significantly inhibited the removal
of CPDs compared with control cells or cells treated with the negative control B7. Twenty-four
hours after irradiation the residual percentage of CPDs in the cells treated with B9 was 35%
compared to 16% in the untreated cells.

Figure 9. Cell-based assay of NER inhibition. (A) Immunofluorescence images at various time
points to detect presence of CPDs in colorectal cancer (HCT-116 ) cells treated with vehicle only
(control), compound B7 (2 µM; negative control), or compound B9 (2 µM). (B) Normalized
fluorescence intensity of the cells based on quantitation of fluorescence from 100 cells randomly
selected per time point. Fluorescence intensity measurements of cells treated with active
inhibitor B9 were significantly different from the non-inhibitory negative control B7 at time
points from 4 – 24 h post-irradiation (p <0.005, Student’s t test).
2.5 Cytotoxicity profile of inhibitors
Crystal violet viability assays were carried out to assess the cellular toxicity of these compounds
to HCT 116 cells and to identify relatively non-cytotoxic concentrations to be used in
experiments sensitizing cells to UV radiation or cyclophosphamide. As can be seen in Fig. 10,
negative control B7 is the least cytotoxic compound to HCT 116 cells. The cytotoxicity of the hit
F06 is greater than that of B9, especially at concentrations ≥ 5 µM. In the case of F06, about
35% of the cells survived after 3 days’ exposure to F06 at a dose of 5 µM, whereas
approximately 95% of the HCT116 cells survived exposure to B9 at the same dose. As a result,

we selected 2 µM as a maximum concentration for F06, while 4 µM of B9 and B7 was chosen as
a maximum concentration to be used in the following assay.
Figure 10. Crystal violet-based viability assay to assess the cytotoxicity profile of B9 (active
inhibitor), B7 (negative control), and F06 (hit).
2.6 Increased sensitisation to cyclophosphamide and UV radiation
The effectiveness of B9 in sensitising HCT 116 cells to UV radiation and cyclophosphamide was
evaluated. As depicted in Fig. 11A, neither 2 nor 4 µM B7 (negative control) had a significant
sensitisation effect on HCT 116 cells to UV radiation, consistent with the observation that B7
shows minimal in vitro inhibition of ERCC1-XPF endonuclease. In contrast, 2 µM F06
displayed a modest sensitisation capacity with increasing the dose (Fig. 11B), while 2 and 4 µM
B9 significantly reduced the survival of cells irradiated with UV (Fig. 11C). We then measured
the effect of B9 on the survival of cells exposed to the DNA inter-strand cross-linking agent
cyclophosphamide (Fig. 11D). Survival curves clearly indicate that B9 (2 µM) significantly
sensitised HCT 116 cells to cyclophosphamide (at concentrations as low as 50 µM) whereas

there was no significant sensitisation effect with B7 (at a higher dose of 4 µM), which is in
accordance with the UV survival data.
A
B
100
10
1
100
10
1
Control
Control
B7 - 2 µM
B7 - 4 µM
F06-2 µM
0
2
4
6
8
10
0
2
4
6
8
10
UV (J/m2)
UV (J/m2)
C
D
100
10
1
100
10
1
Control
Control
0.1
0.01
B7 - 4 µM
B9 - 2 µM
B9 - 2 µM
B9 - 4 µM
0
2
4
6
8
10
0
50
100
150
200
250
300
UV (J/m2)
Cyclophosphamide (µM)
Figure 11. Sensitization of HCT 116 cells to UV irradiation and cyclophosphamide determined
by clonogenic survival assay. A) Cellular survival after exposure of increasing doses of UV
radiation (254 nm) and incubation with 2 and 4 μM B7. B) Cellular survival after exposure of
increasing doses of UV radiation (254 nm) and incubation with 2 μM F06, and C) Cellular
survival after exposure of increasing doses of UV radiation (254 nm) and incubation with 2 and 4
μM B9. D) Cellular survival following exposure to increasing doses of cyclophosphamide and
incubation with B7 (4 µM) or B9 (2 µM). Cell survival curves (±SEM) are based on three
independent sets of experiments. All values resulting from cells treated with 2 μM B9 were

significantly different (p < 0.005, Student’s test) from the values obtained with the B7-treated
cells.
2.7 Predicted binding mode of B9 to the ERCC1-XPF protein
22
In our previous work
we investigated the binding mode of F06 to XPF using molecular
dynamics simulations as a starting point for rational design of analogues with improved affinity.
23
The virtual screening operation in our recent study
yielded compound B9, where the
piperazine ring of F06 was extended with an N,N-dimethylaminoethyl group, and the acridine-
methoxy group (a hydrogen bond acceptor not engaged in any interaction) was replaced by a
hydroxyl group, a hydrogen bond donor. MD simulations indicated that this change resulted in
the creation of an additional hydrogen bond with the side chain of the buried residue V859 (Fig.
12). These features contributed to lowering the binding energy and increased the ligand
efficiency of B9 as compared to F06 23.

Figure 12. Lowest potential energy binding mode of B9 to the XPF HhH2 domain, extracted
from MD simulation (hydrogen bonds are represented in red dashed lines). Modification of the
methoxy group of the parent compound F06 to a hydroxyl group yields an additional hydrogen
bond with the V859 residue, buried in the binding pocket.
2.8 Pharmacokinetic properties of B9 and F06
Preliminary pharmacokinetic (PK) properties of B9 were compared to those of the parent hit
compound F06 22. The log D value of B9 (2.01) was lower than that of F06 (3.86) at pH 7.4.
Exposure of B9 to human liver microsomes revealed that it has a moderate rate of metabolism
while compound F06 displays a more rapid metabolism and high clearance from the liver. Both
compounds reveal an efflux ratio greater than 2 (B9 = >43.39 and F06 = 11.18).
3. Conclusion
The ERCC1−XPF heterodimer is a 5′−3′ structure-specific endonuclease that has a central role in
several important DNA repair pathways, especially NER and ICL. It is required to protect the
integrity of the genome as well as to prevent damage-induced mutations by different insults;
however, it hampers the efficacy of genotoxic cancer therapeutics such as radiotherapy or
chemotherapy with cross-linking agents such as platinum containing compounds. Inhibiting
DNA repair systems in cancer cells using small molecules is a promising approach to potentiate
the effect of such therapies. In our previous study, we employed a computational drug design
approach to rationally design novel compounds based on the hit F06, resulting in a
pharmacophore model that revealed further possible structural optimisation to modulate the
chemical space of different sites on F06. Herein, we further explored modifications on the
suggested sites on F06, which yielded three series of compounds, and dissected their structure

activity relationships with regard to their ability to inhibit the ERCC1-XPF heterodimer and
sensitise colorectal cancer cells to either UV radiation or cyclophosphamide. All the designed
compounds were synthesized via a convergent and reliable sequence. The inhibitory effects of all
compounds on the full-length ERCC1-XPF protein activity were assessed in vitro. These
experiments identified B9 as a potent inhibitor of ERCC1-XPF nuclease activity, with an IC50
value of 0.49 ± 0.04 μM compared to 1.86 ± 0.25 μM for F06. This result suggests that installing
a polar side-chain on the solvent-exposed region of the piperazine and retaining a large,
polycyclic hydrophobic moiety (acridine) is important for optimized activity, which is in accord
with the results of the effect of Gen C and D on the protein complex. Installing a hydrogen bond
donor group on the acridine, as in B5 or B9, increased the inhibitory effect in comparison to the
hydrogen bond acceptor methoxy group such as in F06. Docking studies indicated that this
change allowed an additional hydrogen bond interaction with the side chain of the buried V859
residue the XPF binding pocket. The binding constant for B9 with ERCC1-XPF was
experimentally measured as 85 nM, as compared with a Kd value of 140 nM of F06. The fact that
B7 (negative control) and B9 show disparity in their activity and binding profiles on ERCC1-
XPF despite both compounds possessing an acridine group also provides a strong evidence that
the protein complex inhibition does not involve simple DNA intercalation by the acridine. B9
exhibited inhibition of removal of CPDs formed upon UV irradiation, and sensitized the HCT
116 cells to cyclophosphamide- and UV- induced cytotoxicity, and indication of inhibition of
ICL and NER. Comparison of the partial pharmacokinetic evaluation results of B9 and F06
indicates that B9 has more favorable properties than the F06 particularly by showing a moderate
clearance from the liver, a better permeability, and a lower log D. Since hydrogen bond donor
group installation on the acridine moiety enhances the activity, we intend to explore a wider

range of hydrogen bond donor groups to be installed on the acridine moiety and contrast their
ERCC1-XPF inhibitory activity with B9. In summary, the analysis of the SAR studies identified
B9, which can be combined with other existing DNA-damaging therapies to synergise their
effects by sensitizing cancer cells through inhibition of the ERCC1-XPF DNA repair machinery.
4. Experimental section
4.1 Synthesis of inhibitors
4.1.1 F06 (1)
The hit compound F06 (1) has been synthesized according to our previously reported
procedures22.
4.1.2 Synthesis of Gen B compounds
4.1.2.1 6,9-dichloroacridin-2-ol (Gen B2)
29
Synthesis of B1 was achieved by the reported procedures but with minor modifications; a
mixture of the commercially available 6,9-dichloro-methoxyacridine, B3, (0.28 g, 1.0 mmol) and
1.0 M BBr3 in DCM solvent (2.87 mL, 2.87 mmol) was placed in 10 mL round bottom flask
under nitrogen atmosphere. The reaction mixture was stirred at -78 °C for 20 min and then
warmed to rt and stirred overnight. Upon complete consumption of starting material (monitored
by TLC; 20% MeOH/DCM), the reaction was quenched by adding satd NaHCO3 until formation
of a yellowish solid. The solid residue was filtered and washed with H2O and satd NaHCO3 three
times followed by drying overnight under high vacuum to yield a pure yellow solid (B1).
Spectral data were in agreement with those previously reported. 29
4.1.2.2 4-amino-2-((4-methylpiperazin-1-yl)methyl)phenol (B4)
B4 was synthesized according to the reported procedures 30; stoichiometric amounts of N-methyl

piperazine (0.12 mL, 1.0 mmol), p-acetaminophenol (0.15 g, 1.0 mmol) and 37% formaldehyde
(0.10 mL, 5 mmol) afforded B4 as brownish oil. Spectral data were in agreement with those
previously reported. 30
4.1.2.3 6-chloro-9-((4-hydroxy-3-((4-methylpiperazin-1-yl)methyl)phenyl)amino)acridin-2-ol
(B5)
The procedure employed for the synthesis of F06 synthesis was used to synthesize B5 with the
following reactants: N-methyl piperazine (0.12 mL, 1.0 mmol), acetaminophen (0.15 g, 1.0
mmol), 37% formaldehyde (0.10 mL, 5 mmol), and 6,9-dichloro-2-hydroxyacridine B2 (0.26 g,
1.0 mmol) to afford the product as orange-red semisolid (0.25 g) in 56 % yield; Rf 0.40 (1:9,
MeOH:DCM); IR (cast film) νmax = 3266, 2923, 1628, 1572, 1496, 1250, 1138, 1002, 929, 826,
778 cm-1; 1H NMR (500 MHz, CD3OD) δ 8.02 (d, J = 9.3 Hz, 1H), 7.88 (d, J = 2.1 Hz, 1H), 7.83
(d, J = 9.3 Hz, 1H), 7.36 (d, J = 9.3 Hz, 1H), 7.24 (d, J = 2.6 Hz, 1H), 7.18 (dd, J = 9.3, 2.1 Hz,
1H), 6.84 (dd, J = 8.6, 1.8 Hz, 1H), 6.71 (d, J = 8.6 Hz, 1H), 6.64 (d, J = 2.6 Hz, 1H), 3.54 (s,
13
2H), 2.88 – 2.29 (m, 8H), 2.27 (s, 3H). OH and NH protons were not observed; C NMR (126
MHz, CD3OD) δ 155.0, 153.9, 148.2, 147.2, 146.1, 138.7, 136.4, 129.7, 127.2, 126.3, 124.9,
123.8, 122.1, 121.4, 121.3, 118.2, 117.3, 105.9, 60.4, 55.8 (2C), 53.0 (2C), 49.0, 45.8; HRMS
(ESI) calcd for C25H26ClN4O2 [M + H]+ 449.1739; found 449.1739. HERE
4.1.2.4 6-chloro-9-((4-hydroxy-3-((4-methylpiperazin-1-yl)methyl)phenyl)amino)acridin-2-yl
benzoate (B6)
B5 (0.05 g, 0.1 mmol) was stirred with benzoyl chloride (0.01 mL, 0.1 mmol) in 3mL of
anhydrous solvent mixture DCM/Et3N (20:1) for 1 hr at rt. The reaction mixture was diluted with
water and the aqueous layer was extracted with DCM. The combined organic layers were then
washed with brine, concentrated under vacuum, and purified by column chromatography

(gradient elution with 2% – 15% MeOH:DCM system) to afford product as orange-redd
semisolid (0.04 g) in 74 % yield; Rf 0.45 (1:9, MeOH:DCM); IR (cast film) νmax = 3268, 2926,
1692, 1628, 1582, 1490, 1238, 1135, 1002, 923, 828, 772 cm-1; 1H NMR (500 MHz, CD3OD) δ
8.09 – 8.05 (m, 2H), 7.88 (d, J = 9.2 Hz, 1H), 7.73 – 7.62 (m, 4H), 7.54 – 7.47 (m, 3H), 7.04 (dd,
J = 9.2, 2.1 Hz, 1H), 6.83 (dd, J = 8.5, 2.6 Hz, 1H), 6.73 (d, J = 8.5 Hz, 1H), 6.70 (d, J = 2.6 Hz,
1H), 3.55 (s, 2H), 2.70 – 2.29 (m, 8H), 2.26 (s, 3H). OH and NH protons were not observed; 13C
NMR (126 MHz, CD3OD) δ 166.5, 155.1, 151.2, 147.4, 146.9, 145.5, 139.5, 138.1, 135.1, 131.0
(2C), 130.4, 129.9 (2C), 128.3, 128.1, 126.2, 124.1, 124.0, 123.4, 123.2, 122.8, 118.7, 118.0,
117.7, 116.9, 60.3, 55.5 (2C), 52.8 (2C), 45.6; HRMS (ESI) calcd for C32H30ClN4O3 [M + H]+
553.2001; found 553.1995.
4.1.2.5 4-((2-(benzyloxy)-6-chloroacridin-9-yl)amino)-2-((4-methylpiperazin-1-yl)methyl)phenol
(B7)
A suspension of NaH (60 % dispersion in mineral oil 0.15 g, 0.4 mmol) in DMF (2 mL) was
added dropwise to a stirred solution of B5 (0.08 g, 0.2 mmol) in DMF (2 mL) and stirred for
10 min prior to the addition of benzyl bromide (0.05 mL, 0.4 mmol). After 1.5 h stirring, the
reaction mixture was concentrated under vacuum and the residue was diluted with DCM prior to
washing with saturated solution of NH4Cl. The aqueous layer was extracted with DCM, and the
combined organic layers were washed with brine. After concentration under vacuum, the mixture
was purified by column chromatography (gradient elution with 3% – 10% MeOH:DCM system)
to afford the product as light orange-red semisolid (0.03 g) in 30 % yield; Rf 0.40 (1:9,
MeOH:DCM); IR (cast film) νmax = 3268, 2926, 1692, 1628, 1582, 1490, 1238, 1135, 1002, 923,
828, 772 cm-1; 1H NMR (500 MHz, CD3OD) δ 8.10 (d, J = 9.3 Hz, 1H), 7.88 (d, J = 2.1 Hz, 1H),
7.83 (d, J = 9.3 Hz, 1H), 7.45 (dd, J = 9.4, 2.7 Hz, 1H), 7.39 – 7.29 (m, 6H), 7.24 (dd, J = 9.3,