Rad51/BRCA2 disruptors inhibit homologous recombination and synergize with olaparib in pancreatic cancer cells

Article abstract of DOI:10.1016/j.ejmech.2019.01.008

Olaparib is a PARP inhibitor (PARPi). For patients bearing BRCA1 or BRCA2 mutations, olaparib is approved to treat ovarian cancer and in clinical trials to treat breast and pancreatic cancers. In BRCA2-defective patients, PARPi inhibits DNA single-strand break repair, while BRCA2 mutations hamper double-strand break repair. Recently, we identified a series of triazole derivatives that mimic BRCA2 mutations by disrupting the Rad51-BRCA2 interaction and thus double-strand break repair. Here, we have computationally designed, synthesized, and tested over 40 novel derivatives. Additionally, we designed and conducted novel biological assays to characterize how they disrupt the Rad51-BRCA2 interaction and inhibit double-strand break repair. These compounds synergized with olaparib to target pancreatic cancer cells with functional BRCA2. This supports the idea that small organic molecules can mimic genetic mutations to improve the profile of anticancer drugs for precision medicine. Moreover, this paradigm could be exploited in other genetic pathways to discover innovative anticancer targets and drug candidates.

Full text of DOI:10.1016/j.ejmech.2019.01.008

Accepted Manuscript
Rad51/Brca2 disruptors inhibit homologous recombination and synergize with
olaparib in pancreatic cancer cells
Marinella Roberti, Fabrizio Schipani, Greta Bagnolini, Domenico Milano, Elisa
Giacomini, Federico Falchi, Andrea Balboni, Marcella Manerba, Fulvia Farabegoli,
Francesca De Franco, Janet Robertson, Saverio Minucci, Isabella Pallavicini,
Giuseppina Di Stefano, Stefania Girotto, Roberto Pellicciari, Andrea Cavalli
PII:
S0223-5234(19)30008-X
DOI:
https://doi.org/10.1016/j.ejmech.2019.01.008
EJMECH 11015
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 19 October 2018
Revised Date: 3 January 2019
Accepted Date: 4 January 2019
Please cite this article as: M. Roberti, F. Schipani, G. Bagnolini, D. Milano, E. Giacomini, F. Falchi, A.
Balboni, M. Manerba, F. Farabegoli, F. De Franco, J. Robertson, S. Minucci, I. Pallavicini, G. Di Stefano,
S. Girotto, R. Pellicciari, A. Cavalli, Rad51/Brca2 disruptors inhibit homologous recombination and
synergize with olaparib in pancreatic cancer cells, European Journal of Medicinal Chemistry (2019), doi:
https://doi.org/10.1016/j.ejmech.2019.01.008.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Rad51/Brca2 Disruptors Inhibit Homologous Recombination and Synergize with
Olaparib in Pancreatic Cancer Cells
Marinella Roberti,1 Fabrizio Schipani,2 Greta Bagnolini,1,2 Domenico Milano,2 Elisa Giacomini,2
Federico Falchi,2,§ Andrea Balboni,2,3 Marcella Manerba,3 Fulvia Farabegoli,1 Francesca De
Franco,4 Janet Robertson,4 Saverio Minucci,5,6 Isabella Pallavicini,5 Giuseppina Di Stefano,3
Stefania Girotto,2 Roberto Pellicciari,4 and Andrea Cavalli*1,2
1Department of Pharmacy and Biotechnology, University of Bologna, Via Belmeloro 6, 40126
Bologna, Italy.
2Computational & Chemical Biology, Istituto Italiano di Tecnologia, via Morego 30, 16163
Genova, Italy.
3Department of Experimental, Diagnostic and Specialty Medicine, University of Bologna, Via S.
Giacomo 14, 40126 Bologna, Italy.
4TES Pharma S.r.l., Via Palmiro Togliatti 22bis, I-06073 Loc. Terrioli, Corciano, Perugia, Italy.
5Department of Experimental Oncology at the European Institute of Oncology, IFOM-IEO Campus,
Via Adamello 16, 20100 Milan.
6Department of Biosciences, University of Milan, Via Celoria 26, 20100 Milan.
§
Present address: Molecular Horizon srl, Via Montelino 30, 06084 Bettona (PG), Italy.
*Corresponding Author: Department of Pharmacy and Biotechnology, University of Bologna, Via
Belmeloro 6, 40126 Bologna, Italy.
Computational & Chemical Biology, Istituto Italiano di Tecnologia, via Morego 30, 16163 Genova,
Italy. E-mail address: andrea.cavalli@iit.it
1

ACCEPTED MANUSCRIPT
Abstract
Olaparib is a PARP inhibitor (PARPi). For patients bearing BRCA1 or BRCA2 mutations, olaparib
is approved to treat ovarian cancer and in clinical trials to treat breast and pancreatic cancers. In
BRCA2-defective patients, PARPi inhibits DNA single-strand break repair, while BRCA2
mutations hamper double-strand break repair. Recently, we identified a series of triazole derivatives
that mimic BRCA2 mutations by disrupting the Rad51-BRCA2 interaction and thus double-strand
break repair. Here, we have computationally designed, synthesized, and tested over 40 novel
derivatives. Additionally, we designed and conducted novel biological assays to characterize how
they disrupt the Rad51-BRCA2 interaction and inhibit double-strand break repair. These
compounds synergized with olaparib to target pancreatic cancer cells with functional BRCA2. This
supports the idea that small organic molecules can mimic genetic mutations to improve the profile
of anticancer drugs for precision medicine. Moreover, this paradigm could be exploited in other
genetic pathways to discover innovative anticancer targets and drug candidates.
Keywords: synthetic lethality, homologous recombination, protein-protein small molecule
inhibitors, anticancer drugs, PARP inhibitors.
2

ACCEPTED MANUSCRIPT
1. Introduction
In the past two decades, cancer research has focused on discovering tumor-specific signaling and
metabolic pathways that might be exploited for pathway-targeted anticancer therapies [1,2]. To
date, regulators have approved several small molecules and monoclonal antibodies, which directly
inhibit genes and proteins that are critical in oncogenic signal networks. These have increased
survival in patients with previously intractable cancers [3,4,5,6,7,8]. However, the first generation
of targeted therapies have some limitations, including the development of resistance and on- and
off-target toxicities [9,10]. Therefore, researchers are increasingly turning to precision medicine,
using genetic changes in a patient’s tumor to determine their treatment.
Synthetic lethality is a new concept with great potential for discovering new anticancer
molecules within the framework of precision medicine. Synthetic lethality arose from genetic
studies in model organisms. Two genes are synthetically lethal if the perturbation of either gene
alone is compatible with cell viability, but simultaneous perturbation of both genes leads to cell
death. In principle, it is possible to use a small molecule to target the synthetically lethal partner of
an altered gene in cancer cells but not in normal cells. This should selectively kill cancer cells while
sparing normal cells [11,12,13,14].
One example of synthetic lethality is the use of poly [ADP-ribose] polimerase (PARP) inhibitors
in oncology patients with BRCA1/2 mutations (Fig. 1A). BRCA2 is important for repairing DNA
double-strands breaks (DSBs) by homologous recombination (HR), whereas PARP is important for
repairing single-strand breaks. Synthetic lethality arises from the simultaneous impairment of the
repair mechanisms for single-strand and double-strand breaks. Developed by AstraZeneca, olaparib
(Lynparza™, Fig. 1A) was the first PARP inhibitor (PARPi) approved to treat advanced ovarian
cancer associated with defective BRCA genes [15]. Several clinical trials are now studying olaparib
to treat breast and pancreatic tumors associated with germline BRCA mutations [16].
One of BRCA2’s key mechanisms is to recruit Rad51, an evolutionarily conserved recombinase
that uses HR to repair double-strand breaks [17]. In parallel, Rad51 overexpression has been
3

ACCEPTED MANUSCRIPT
reported in a wide variety of cancers (e.g. breast, pancreatic), where enhanced HR activity has been
observed [18]. Rad51 and BRCA2 interact through eight well-conserved motifs, namely the BRC
repeats [19, 20]. The X-ray crystallographic structure of BRC4 (the fourth BRC repeat) is available
in complex with the catalytic domain of Rad51 [21]. This makes the Rad51-BRCA2 interaction
suitable for the structure-based design of small molecule disruptors. Indeed Lee et al. recently
identified a small molecule inhibitor of Rad51-BRCA2 for potential cancer treatment [22, 23].
In this context, we recently proposed a new anticancer drug discovery concept, which combines
Rad51-BRCA2 disruptors with olaparib to mimic the synthetic lethality described above (Fig. 1B)
[24]. Our working hypothesis was that, by disrupting the Rad51-BRCA2 interaction, we could
make cancer cells more sensitive to PARPi, DNA-damaging agents, and radiotherapy. In particular,
we reasoned that, by administering Rad51-BRCA2 disruptors to individuals without BRCA
mutations, it would be possible to mimic the enhanced sensitivity to PARPi observed in BRCA2-
defective oncology patients. Due to their increased genetic instability, cancer cells experience DNA
damage more frequently than healthy cells. They should therefore be preferentially affected by the
inhibition of DNA repair produced by the synergy of olaparib and the Rad51-BRCA2 disruptors.
4

ACCEPTED MANUSCRIPT
Figure 1. a) Mechanism of synthetic lethal death induced by the PARP inhibitor olaparib in BRCA-defective cancer
cells. b) Mechanism of synthetic lethal death triggered by a combination of olaparib and BRCA2-RAD51 disruptors. In
this case, synthetic lethality is fully triggered by small organic molecules.
We first carried out a virtual ligand screening campaign using commercial databases (Asinex and
Life Chemical databases; overall more than 1.5 millions of compounds), which allowed us to
identify potential triazole-based hit compounds. Subsequent ELISA biochemical assay showed that
1 (Fig. 2) was able to disrupt the Rad51-BRCA2 interaction. Preliminary modification of the linker
connecting the triazole and the phenyl ring led to the propyl phenyl derivative 2, which was studied
at the cellular level. In line with our hypothesis, 2 increased the sensitivity to olaparib in cancer
cells BxPC-3 with fully functional BRCA2 [24]. This synergistic effect was not observed in Capan-
1, pancreas adenocarcinoma cells that lack functional BRCA2.
Figure 2. Structure of the previously identified 1 and 2.
To discover more effective compounds, we here design and synthesize a library of new triazole
analogues (3-42, Table 1-2). In particular, we attempt to depict general structure-activity
relationships (SAR) of 3−42 in inhibiting the Rad51-BRCA2 interaction. Additionally, we also
5

ACCEPTED MANUSCRIPT
report on novel biological investigations to further elucidate the mechanistic aspects of the
synergistic activity of Rad51-BRCA2 disruptors and PARPi.
2. Results and discussion
2.1 Design approach
Targeting protein-protein interactions (PPI) is an attractive strategy for designing innovative
drugs for highly complex diseases, such as cancer [25]. However, it is rather difficult to identify and
optimize PPI disruptors. This is because proteins usually interact with other proteins via shallow
surface sites. This makes it difficult to optimize ligands using classic structure-based approaches.
To improve binding at surface sites, many weak interactions must be optimized. In-depth studies of
the target structure may identify surface pockets as hot spots to be targeted. We identified a
potentially suitable pocket, often referred to as the “FxxA pocket”. This pocket is one of the two
Rad51 regions where BRC4 (i.e. the fourth BRC repeat responsible for binding of BRCA2 to
Rad51) [24] binds its counterpart. This structural analysis prompted us to run a virtual screening
campaign based on high-throughput docking at the FxxA pocket in order to identify Rad51-BRCA2
disruptors. With this strategy, we discovered a series of triazole-based compounds. 1 and 2 were
selected as initial hit candidates [24]. Notably, triazoles bear a peculiar structure that makes them
particularly suited to disrupting protein-protein interactions [26,27,28]. The binding mode of 1 to
Rad51 (Fig. 3a), as obtained by docking simulations, shows noticeable similarities to the
crystallographic structure of Rad51 in complex with BRC4. Specifically, the docking model
suggests that (i) the phenyl ring of the ethyl phenyl group of 1 binds (similarly to the Phe1524 of
BRC4) into the hydrophobic pocket outlined by the side chains of Met158, Ile160, Ala190, Leu203,
Ala207, and Met210 of Rad51; (ii) the carbonyl group of the benzothiazolone moiety forms
hydrogen bonds with the sidechains of Rad51, His199, and Gln206, similarly to the backbone
carbonyl of Leu1522 of BRC4. Likewise, (iii) the acetamide carbonyl of 1 binds the backbone
Tyr191 of Rad51, similarly to His1525 of BRC4.
6

ACCEPTED MANUSCRIPT
To improve the protein-protein inhibitory activity of 1, we conducted a chemical modification
campaign around the triazole moiety. Figure 3b provides an overview of the structural variants
introduced on the triazole scaffold.
First, several different acetamide chains were introduced in position 3 (blue region, Fig. 3b)
bearing different heterocycles, saturated or unsaturated rings, leaving the carbonyl group unchanged
(3-10, Table 1). Additionally, to investigate its role, the acetamide function was removed (11, Table
1). Regarding position 4 (red region, Fig. 3b), the first set of structural changes involved
introducing different substituents (namely methyl, hydroxyl, and fluorine) on the phenyl ring, while
retaining the two carbons ethyl chain (12-18, Table 1). In position 5 (green region, Fig. 3b), the
methylene benzothiazolone moiety was removed or replaced with other aromatic rings to probe its
role (19-22, Table 1). Moreover, the methylene benzothiazolone moiety was modified by
introducing different substituents on the aromatic ring, including hydroxy and methoxy groups and
a fluorine atom (23-25, Table 1).
Based on the promising biological activities of the 4-N-propyl phenyl triazole derivative 2, a
second set of compounds was envisaged (26-42). Similarly to compound 1, modifications were
made to the moieties in position 3 (26-32) and 5 (33-39) of the triazole ring, leaving the linear
propyl chain in position 4 unchanged. Moreover, the phenyl fragment of the propyl phenyl region in
position 4 was substituted by 2, 3 and 4 pyridine moieties (40-42) (Fig. 3b).
B
A
7

ACCEPTED MANUSCRIPT
Figure 3. a) Compound 1 docked into the FxxA domain of Rad51. The small molecule recapitulates major interactions
the peptide BRC4 establishes with the biological target. b) Overview of the optimization strategy of 1 for SAR
exploration. 2 was investigated in a similar way to 1, leaving the linear propyl chain in position 4 unchanged.
2.2 Chemistry
Scheme 1 illustrates the common synthetic strategy for achieving most of the desired compounds
(3-21, 23, 25-36, 38-42). Hydrazides 43a-n and the appropriate isothiocyanates 44a-l underwent
cyclocondensation in the presence of triethylamine to give 1,2,4-triazole-3-thiol intermediates 45-
69. S-alkylation of 45-55, 57-63, and 65-69 with 2-chloro-acetamides 70a-m led to final
compounds 3-21, 23, 25-36, and 38-42. The 5-methoxybenzothiazolone derivative 23 was in turn
treated with BBr3 to obtain the corresponding hydroxy analog 24 (see Supporting Information
Scheme S1). Compounds 22 and 37 were obtained from the Boc-protected 1,2,4-triazole-3-thiol
intermediates 56 and 64, which were alkylated with 2-chloro-N-cyclopentylacetamide 70l to afford
the corresponding derivatives 71-72. In turn, 71-72 were Boc-deprotected to afford the final 22, 37
respectively (see Supporting Information Scheme S2). The hydrazides 43a-j, not commercially
available, were obtained by reaction of hydrate NH2NH2 with the corresponding methyl, ethyl, or
benzyl ester 73a-j (Scheme 2). The esters 73a-j were obtained by nucleophilic substitution reaction
between the appropriate amine 74a-i and benzyl, ethyl, or methyl bromoacetate 75, 76, and 77
respectively. In addition to the desired 73i, the reaction between the substituted pyridinone 74i and
methyl bromoacetate 77 unexpectedly provided the substituted pyridine methyl 2-oxiacetate 73j
(Scheme 2, entry 9). 73i and 73j were separated and treated with hydrate NH2NH2 to afford the
corresponding hydrazides 43i and 43j. The hydrazide 43k, which is not commercially available,
was prepared from the Boc-aminomethyl ester 78 according to general procedure. 78 was obtained
by the esterification of aminomethyl benzoic acid 79, which was refluxed with ethanol in the
presence of HCl to achieve the amino ester 80. In turn, 80 was protected with tert-butyloxycarbonyl
(Boc) group to afford 78 (see Supporting Information Scheme S3). The isothiocyanates 44a-j were
8

ACCEPTED MANUSCRIPT
not commercially available and were obtained by reaction of the appropriate amine 81a-j with 1,1-
thiocarbonyldimidazole 82 (Scheme 3). The chloroacetamides 70a-h, not commercially available,
were obtained through nucleophilic substitution reaction between the appropriate amine 83a-h and
chloroacetyl chloride 84 (Scheme 4).
9

ACCEPTED MANUSCRIPT
Scheme 1. General synthesis of final compounds 3-42
10

ACCEPTED MANUSCRIPT
Scheme 2. Synthesis of hydrazides 43a-j
Entry
N°
A
B
C
D
1
2
3
74a
74b
74c
73a
43a
75
75
75
73b
43b
73c
43c
4
75
73d
43d
74d
74e
5
6
75
76
73e
73f
43e
43f
74f
11

ACCEPTED MANUSCRIPT
7
8
74g
74h
77
77
73g
73h
43g
43h
73i
43i
43j
9
77
74i
73j
12

ACCEPTED MANUSCRIPT
Scheme 3. Synthesis of isothiocyanates 44a-j
Scheme 4. Synthesis of chloroacetamides 70a-h
13

ACCEPTED MANUSCRIPT
2.3 Biological evaluation
Our working hypothesis is that small molecule disruptors of Rad51-BRCA2 could synergize
with PARPi to treat ovarian, breast, and pancreatic cancers. Here, we focused on pancreatic cancer
because it is an unmet oncological need. To investigate the mechanism of action of the new triazole
derivatives, different biological assays were performed. First, the ability of compounds 3-42 to
inhibit Rad51-BRCA2 interaction was investigated with a competitive biochemical ELISA assay
against parent compounds 1 and 2 (Table 1-2). This assay is effective in evaluating the ability of
new molecules to compete with BRC4 to bind to Rad51 [24].
For the 4-N-ethyl phenyl triazole series (3-25, Table 1), all compounds were inactive, with the
exception of 3-thio-N-phenylacetamide 3 and 4-N(2-methyphenethyl) 16 derivatives, which were
active in the high micromolar range. These data indicate that no improvement in Rad51-BRCA2
inhibitory activity was achieved with this new series of compounds relative to the parent compound,
1 (EC50 = 53±3 ꢀM).
For the series of the 4-N-propyl phenyl triazoles (26-42, Table 2), 3-thio-N-cyclohexylacetamide
26 showed fairly good potency with an EC50 of 8±2 ꢀM, three-fold higher than the parent 3-thio-N-
cyclopentylacetamide 2 (EC50 = 25±2 ꢀM). The binding mode of these two compounds, obtained by
docking simulations, is reported in the Supporting Information (Fig. S1). 27 and 28 were active in
the micromolar range (EC50 of 24±5 ꢀM and 42±2, respectively), both very similar to the initial hit.
Therefore, activity is marginally affected by replacing the cyclopentyl ring in the 3-thio-N-
cyclopentylacetamide chain with different cycloalkyl rings such as cyclohexyl (26), cyclopropyl
(27), and cyclobutyl (28). Replacing the methylene benzothiazolone moiety yielded compounds that
had reduced inhibitory activity (34, EC50 = 90±5 ꢀM and 39 EC50 = 33±7 ꢀM) or were fully
inactive (33, 35-38). For compounds 40, 41, and 42, the phenyl of the propyl phenyl group was
replaced by a 2, 3, or 4-pyridine ring, respectively. These compounds were inactive. As expected,
PPI disruptors are very difficult to obtain and optimize. Moreover, it is challenging to properly
depict meaningful SAR. Indeed, slight modifications of active compounds led on the one hand to
14

ACCEPTED MANUSCRIPT
molecules that were either fully inactive or with greatly reduced inhibitory activity, while on the
other hand to molecules with improved potency. However, further modifying compounds in that
direction did not necessarily provide reasonable SAR. Protein-protein disruptors usually bind the
target at shallow surface sites, and potency against the target is usually dictated by many weak
interactions. Therefore, it is challenging to identify the optimal pool of substituents that a small
molecule should bear in order to recapitulate the protein-protein interactions. Despite this, we
slightly improved the biological profile of the initial compound (2), as per the biological
investigations reported below.
Compounds 26, 27, 28, and 39 were submitted to cell-based studies. 28 and 39 were discarded
because of poor solubility. 26 and 27 were characterized in human pancreatic carcinoma, using
BxPC3 cells. These cells were chosen for their clinical relevance and for the possibility of using
Capan-1 cells, [29] which lack functional BRCA2 protein and could therefore serve as negative
control. Additionally, BxPC3 and Capan-1 match well in terms of tumor type and differentiation
state [29].
The Rad51-BRCA2 interaction is crucial in recruiting the recombinase Rad51 to cell nuclei,
where it uses homologous recombination (HR) to repair double-strand breaks. We therefore studied
HR inhibition in BxPC3 and Capan-1 cells. The assay is based on cell transfection with two
plasmids bearing the LacZ sequence, each with a different mutation. In transfected cells, the
recombination machinery restores the corrected LacZ sequence, which can be detected by PCR.
Using two different primer sets, both the plasmid couple and the recombination product can be
measured, allowing an estimation of HR efficiency. This assay was applied to cells treated (16 h)
with 2 and with the two new triazole structures (26 and 27) as shown in Figure 4A.
Unexpectedly, 2 increased the HR rate in treated cells, presumably by causing direct DNA
damage via off-target activities. Conversely, 26 and 27 reduced HR in treated BxPC3 cells
suggesting their potential ability to disrupt Rad51-BRCA2 interaction in living cells too. Of the two
molecules, 26 had the strongest effect, with a statistically significant 40% inhibition at 20 µM. 27
15

ACCEPTED MANUSCRIPT
showed a less marked effect, with a 24% inhibition at 40 µM. The limited solubility of 27 further
hindered a dose escalation with this compound. Neither 26 nor 27 inhibited HR in Capan-1 cells
(see Supporting Information, Figure S2), which lack functional BRCA2. This strongly supports
their mechanism of action as being the disruption of the Rad51-BRCA2 interaction. These data also
suggested that 26 and 27 were potentially capable of synergizing with olaparib in pancreatic cancer
cells with functional BRCA2, where olaparib usually is inactive. The two compounds were
therefore tested in BxPC3 and Capan-1 cell viability experiments in combination with olaparib.
Figure 4B reports the results after 72 h exposure to treatments. 27 did not affect olaparib efficacy.
However, 26 significantly increased olaparib efficacy in BxPC3 cells, without affecting its efficacy
in the BRCA2-defective cells (Capan-1). These results further confirm, in a dose-dependent
manner, that HR inhibition might explain the synergistic effect of combining Rad51-BRCA2
disruptors with PARPi in cell lines with fully functional BRCA2, where PARPi are usually inactive.
Further experiments were conducted on 26 to confirm the synergy with olaparib.
16

ACCEPTED MANUSCRIPT
Figure 4. (A) Estimation of homologous recombination efficiency in BxPC-3 cells treated (16 h) with 2 and with the
newly identified BRCA2-Rad51 disruptors 26 and 27. *, p < 0.05 as evaluated by ANOVA followed by Dunnet’s post-
test. (B) Evaluation of cell viability assessed in combination experiments with olaparib (10 µM) + 26 or 27 on BxPC-3
and Capan-1 cultures. The doses showing the best results in the experiments in Fig. 1A were used. Olaparib
concentration was selected on the basis of previous experiments (2). *, p < 0.05 as evaluated by ANOVA followed by
Dunnet’s post-test.
By adopting a longer exposure time to treatments (144 h), we assessed how the 26/olaparib
combination affected cell viability and cell death (Fig. 5A). Cell death was evaluated using the
CellTox Green Dye (Promega) probe, which emits a fluorescence signal only in the presence of
cells with impaired membrane integrity. Again, we observed a statistically significant increase in
olaparib’s power in reducing cell viability. The data of the cell viability experiment plotted in Fig.
17

ACCEPTED MANUSCRIPT
5A were also used to evaluate the combination index (CI) between olaparib and 26. Following the
procedure described previously [24], a CI value of 0.76 ± 0.03 was obtained, which suggests a
potential synergistic effect between the compounds.
However, the enhanced olaparib potency did not produce evidence of increased cell death. An
immunoblotting evaluation of apoptosis markers confirmed this result (See Supporting Information,
Fig. S3). Using the same exposure time to treatments, an immunoblotting evaluation of DNA
damage was also performed by measuring the phosphorylation of H2AX (γ-H2AX), a sensitive and
universal indicator of DNA double-strand break formation [30]. Results are reported in Figure 5B.
The densitometric evaluation of γ-H2AX bands, normalized on actin levels, have been plotted in the
bar graph. It shows that, in cells treated with the combination of olaparib with 26, the DNA damage
signal was >70% higher than that observed in control and in olaparib-treated cells. This experiment
also showed that, when given as a single treatment, 26 did not cause appreciable signs of DNA
damage.
Overall, these data suggest that disrupting Rad51-BRCA2 interaction could be a valuable
strategy for improving olaparib’s antitumor profile in cancer cell lines that are usually refractory to
the drug. This includes pancreatic cancer, which is a major unmet medical need. Additionally, this
study paves the way for the exploitation of the synthetic lethality paradigm in order to discover
innovative anticancer therapies based on other lethal gene pairs, using a similar medicinal chemistry
strategy.
18

ACCEPTED MANUSCRIPT
Figure 5. (A) cell viability and cell death measured in BxPC-3 and Capan-1 cultures after 144 h treatment. 26
significantly increased olaparib’s effect on the viability of BxPC-3 cells. *, p < 0.05 as evaluated by ANOVA followed
by Dunnet’s post-test. No effect on cell death was observed. (B) Immunoblotting evaluation of γ-H2AX. The
combination olaparib + 26 caused a >70% increased signal of DNA damage (see bar graph).
19

ACCEPTED MANUSCRIPT
Table 1. Structures and EC50 of compounds 3-25 on ELISA Assay
Cp R1
3
R2
R3
EC50 (µM)*
95±5
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
100±10
n.a.**
n.a.**
n.a.**
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
n.a.**
n.a.**
20
21
20

ACCEPTED MANUSCRIPT
n.a.**
22
23
n.a.**
n.a.**
n.a.**
24
25
*: All points were tested in triplicate with error bars indicating the standard deviation.
**: not active.
Table 2. Structures and EC50 of compounds 26-42 on ELISA Assay.
EC50(ꢀM)*
Cp
26
R1
R2
R3
8±2
24±5
42±3
27
28
29
30
31
32
n.a.**
n.a.**
n.a.**
n.a.**
n.a.**
33
90±5
n.a.**
n.a.**
34
35
36
n.a.**
37
21

ACCEPTED MANUSCRIPT
n.a.**
33±7
38
39
n.a.**
n.a.**
n.a.**
40
41
42
*: All points were tested in triplicate with error bars indicating the standard deviation.
**: not active.
3. Conclusions
In recent years, innovative anticancer treatments have relied on targeted therapies to improve
drug efficacy and reduce side-effects. In parallel, precision medicine has provided a new paradigm
for identifying individual patients or groups of patients for treatment with tailored pharmacological
medications. For example, PARP inhibitors were recently been approved to treat ovarian cancer in
patients with genetic mutations in the BRCA genes (1 and 2). For patients with BRCA2 mutations,
the underlying molecular mechanism of action is related to the simultaneous blocking of two DNA
repair pathways. Olapirib (the first approved PARPi) chemically inhibits single-strand break repair,
while the BRCA2 germline mutations genetically block the double-strand break repair. This has
been demonstrated in ovarian cancer (for which olaparib was originally approved), and in clinical
trials in breast and pancreatic cancers. Pancreatic cancer is a major unmet medical need. Recently,
we reported a new paradigm dubbed “fully small-molecule-induced synthetic lethality.” This
paradigm was based on the observation that, by using a small organic molecule to disrupt the
Rad51-BRCA2 interaction, it is possible to mimic the effect of BRCA2 germline mutations, thus
widening the population of patients eligible for treatment with olaparib and other PARPi. In the
present study, we further confirmed this concept by investigating the SAR of the previously
discovered triazole derivatives. We also improved the biological screening cascade with
22

ACCEPTED MANUSCRIPT
experiments designed to confirm the molecular mechanism of action (disruption of Rad51-BRCA2
interaction) in cancer cells. We obtained one molecule, 26, with an improved profile relative to the
initial hits (according to a biochemical ELISA essay), and a clear mechanism of action allowing
synergy with olaparib in pancreatic cancer cells (BxPC3), where olaparib is normally inactive. In
summary, 26 i) inhibited the BRCA2-Rad51 interaction; ii) hit its target and inhibited its function in
living cells; iii) significantly increased the therapeutic power of olaparib alone in cells with a
functional BRCA2-Rad51 pathway, where usually olaparib is inactive; and iv) increased the
formation of double-strand breaks when administered in combination with olaparib.
However, we could not fully reproduce the paradigm of synthetic lethality with 26. A possible
explanation is its low-level potency, which did not cause HR inhibition greater than 40%. In
addition, the inherent resistance to apoptosis of BxPC3 cells, which bear a mutant p53 [31], could
have further prevented apoptosis. Therefore, further biological experiments and new classes of
Rad51-BRCA2 disruptors are needed to confirm this paradigm for discovering innovative
anticancer therapies.
We are confident that the synthetic lethality framework will provide new pathways, targets, and
ideas for identifying small molecules with complex and innovative mechanisms of action as next-
generation anticancer therapies.
4. Experimental section
4.1 Chemistry
4.1.1 General Chemical Methods
Solvents and reagents were obtained from commercial suppliers and used without further
purification. If required, solvents were distilled prior to use. For simplicity, solvents and reagents
are indicated as follows: acetonitrile (MeCN), 1,8-Diazabicyclo(5.4.0)undec-7-ene (DBU),
cyclohexane (Cy), dichloromethane (DCM), diethyl ether (Et2O), dimethylsulfoxide (DMSO),
ethanol (EtOH), ethyl acetate (EtOAc), methanol (MeOH), triethylamine (TEA). Thin-layer
23

ACCEPTED MANUSCRIPT
chromatography analyses were performed using pre-coated Supelco silica gel on TLC Al foils 0.2
mm and visualized by UV (254 nm), and/or KMnO4 stain. Automated column chromatography
purifications were done using a Teledyne ISCO apparatus (CombiFlash® Rf) with pre-packed silica
gel columns of different sizes (from 4 g to 120 g). Mixtures of increasing polarity of Cy and EtOAc
or DCM and MeOH were used as eluents. Reactions involving microwave irradiation were
performed using Explorer® -48 positions instrument (CEM). NMR experiments were run on a
Bruker Avance III 400 system (400.13 MHz for 1H, and 100.62 MHz for 13C), equipped with a BBI
probe and Z-gradients. Spectra were acquired at 300 K, using deuterated dimethylsulfoxide
1
13
(DMSO-d6) or deuterated chloroform (Chloroform-d) as solvent. Chemical shifts for H and C
spectra were recorded in parts per million using the residual non-deuterated solvent as the internal
1
13
1
standard (for DMSO-d6: 2.50 ppm, H; 39.52 ppm, C; for Chloroform-d: 7.26 ppm, H; 77.16
13
ppm, C). Data are reported as follows: chemical shift (ppm), multiplicity (indicated as: bs, broad
signal; s, singlet; d, doublet; t, triplet; q, quartet; p, quintet; sx, sextet; m, multiplet and
combinations thereof), coupling constants (J) in Hertz (Hz) and integrated intensity. UPLC/MS
analyses were run on a Waters ACQUITY UPLC/MS system consisting of an SQD (Single
Quadrupole Detector) Mass Spectrometer equipped with an Electrospray Ionization interface and a
Photodiode Array Detector. PDA range was 210-400 nm. Analyses were performed on an
ACQUITY UPLC BEH C18 column (50 x 2.1 mmID, particle size 1.7 ꢀm) with a VanGuard BEH
C18 pre-column (5 x 2.1 mmID, particle size 1.7 ꢀm). Mobile phase was 10 mM NH4OAc in H2O
at pH 5 adjusted with AcOH (A) and 10 mM NH4OAc in CH3CN/H2O (95:5) at pH 5 (B).
Electrospray ionization in positive and negative mode was applied. Analyses were performed with a
gradient: 5 to 95% B over 3 min; flow rate 0.5 mL/min; temperature 40 °C. Compounds were
named using the naming algorithm developed by CambridgeSoft Corporation and used in
ChemBioDraw Ultra 15.0. All final compounds displayed ≥ 98% purity as determined by
UPLC/MS analysis.
24

ACCEPTED MANUSCRIPT
PAINS analysis. Compounds 1-3, 16, 26-28, 34 and 39, which turned out to be biologically active,
were analyzed for known classes of interference compounds. According to the Free ADME-Tox
Filtering Tool (FAFDrugs4) program (http://fafdrugs4.mti.univ-paris-diderot.fr/) none of them was
recognized to cause pan assay interference.
4.1.2 General procedure for synthesis of final triazoles (3-42)
The appropriate 1,2,4-triazole-3-thiol (45-69, 1 equiv), 2-chloro-acetamide (70a-m, 1.1 equiv),
Cs2CO3 (1.1 equiv) and NaI (0.05 equiv) were stirred in dry MeCN at 50 °C, and reaction progress
was monitored by UPLC/MS. Then water was added, and the final compounds 3-42 were purified
by washing the precipitated powder or by flash chromatography (Scheme 1).
Below, we report the characterization of the most active compounds. (See also Supplementary
data).
N-cyclohexyl-2-((5-((2-oxobenzo[d]thiazol-3(2H)-yl)methyl)-4-(3-phenylpropyl)-4H-1,2,4-
triazol-3-yl)thio)acetamide
26.
3-((5-mercapto-4-(3-phenylpropyl)-4H-1,2,4-triazol-3-
yl)methyl)benzo[d]thiazol-2(3H)-one 59 (80 mg, 0.21 mmol), 2-chloro-N-cyclohexylacetamide 70b
(40 mg, 0.23 mmol), Cs2CO3 (75 mg, 0.23 mmol), and a catalytic amount of NaI in 8 mL of dry
MeCN were allowed to react for 4 h according to general procedure. The desired product
precipitated
from
water,
and
3-((5-mercapto-4-(3-phenylpropyl)-4H-1,2,4-triazol-3-
yl)methyl)benzo[d]thiazol-2(3H)-one washed with water, Et2O and dried. Precipitate was purified
by column chromatography, solvent system: cyclohexane/EtOAC. Clean compound eluted with
1
EtOAc 100%, to give 26 (89 mg, 80% yield) as a white solid. H NMR (400 MHz, DMSO-d6) δ
7.99 (d, J = 7.7 Hz, 1H), 7.69 (dd, J = 7.8, 1.2 Hz, 1H), 7.47 (dd, J = 8.2, 1.1 Hz, 1H), 7.36 (td, J =
7.8, 1.3 Hz, 1H), 7.30 – 7.17 (m, 6H), 5.39 (s, 2H), 4.06 (t, 2H), 3.82 (s, 2H), 3.47-3.49 (m, 1H),
13
2.59 (t, 2H), 1.87-1.79 (m, 2H), 1.63-1.58 (m, 4H), 1.23-1.02 (m, 6H) ppm. C NMR (101 MHz,
DMSO-d6) δ 169.1, 165.53, 150.3, 150.2, 140.5, 136.5, 128.3 (2C), 128.1 (2C), 126.6, 126.0, 123.6,
122.9, 121.2, 112.1, 47.9, 43.6, 37.0, 36.9, 32.1 (2C), 32.0, 31.0, 25.1, 24.3 (2C) ppm. UPLC-MS
(ESI, m/z) Rt = 2.45 min – 522 (M+H)+; UPLC-MS purity (UV 215 nm): > 99.5%
25

ACCEPTED MANUSCRIPT
N-cyclopropyl-2-((5-((2-oxobenzo[d]thiazol-3(2H)-yl)methyl)-4-(3-phenylpropyl)-4H-1,2,4-
triazol-3-yl)thio)acetamide
27.
3-((5-mercapto-4-(3-phenylpropyl)-4H-1,2,4-triazol-3-
yl)methyl)benzo[d]thiazol-2(3H)-one 59 (100 mg, 0.26 mmol), 2-chloro-N-cyclopropylacetamide
70m (39 mg, 0.29 mmol), Cs2CO3 (95 mg, 0.29 mmol) and a catalytic amount of NaI in 5 mL of
dry MeCN were allowed to react overnight according to general procedure. The desired product
precipitated from water, and was washed with water, Et2O and dried overnight under vacuum at 60
1
°C to give 27 (120 mg, quantitative yield) as a white powder. H NMR (400 MHz, DMSO-d6) δ
8.23 (d, J = 4.1 Hz, 1H), 7.70 (dd, J = 7.8, 1.2 Hz, 1H), 7.48-7.46 (m, 2H), 7.36 (td, J = 7.8, 1.3 Hz,
1H), 7.30 – 7.17 (m, 7H), 5.39 (s, 2H), 4.05 (t, J = 7.9 Hz, 2H), 3.81 (s, 2H), 2.61 – 2.53 (m, 3H),
1.87-1.79 (m, 2H), 0.59-0.54 (m, 2H), 0.34 – 0.30 (m, 2H) ppm. 13C NMR (101 MHz, DMSO-d6) δ
169.1, 167.6, 150.3, 150.2, 140.5, 136.5, 128.4 (2C), 128.1 (2C), 126.6, 126.0, 123.6, 122.9, 121.2,
112.1 , 43.6, 36.9, 36.5, 32.0, 31.0, 22.5, 5.5 (2C) ppm. UPLC-MS (ESI, m/z) Rt = 2.15 min – 480
(M+H)+; UPLC-MS purity (UV 215 nm): > 99.5%.
N-cyclobutyl-2-((5-((2-oxobenzo[d]thiazol-3(2H)-yl)methyl)-4-(3-phenylpropyl)-4H-1,2,4-
triazol-3-yl)thio)acetamide
28.
3-((5-mercapto-4-(3-phenylpropyl)-4H-1,2,4-triazol-3-
yl)methyl)benzo[d]thiazol-2(3H)-one 59 (100 mg, 0.26 mmol), 2-chloro-N-cyclobutylacetamide
70a (43 mg, 0.29 mmol), Cs2CO3 (95 mg, 0.29 mmol) and a catalytic amount of NaI in 5 mL of dry
MeCN were allowed to react overnight according to general procedure. The desired product
precipitated from water, and was washed with water, Et2O and dried. Precipitate was purified by
column chromatography, solvent system: DCM/MeOH. Clean compound eluted with MeOH 4%, to
give 28 (102 mg, 79% yield) as a white solid. 1H NMR (400 MHz, DMSO-d6) δ 8.39 (d, J = 7.6 Hz,
1H), 7.69 (dd, J = 7.8, 1.2 Hz, 1H), 7.47 (dd, J = 8.1, 1.1 Hz, 1H), 7.41 – 7.29 (m, 1H), 7.32 – 7.16
(m, 6H), 5.39 (s, 2H), 4.14 – 4.07 (m, 1H), 4.06- 4.03 (m, 2H), 3.82 (s, 2H), 2.62 – 2.55 (m, 2H),
2.13 – 1.98 (m, 2H), 1.90 – 1.69 (m, 4H), 1.64 – 1.50 (m, 2H) ppm. 13C NMR (101 MHz, DMSO) δ
169.1, 165.3, 150.3, 140.5,136.6, 128.4 (2C), 128.1 (2C), 126.6, 126.0, 123.6, 122.9, 121.2, 112.1,
26

ACCEPTED MANUSCRIPT
99.5, 43.6, 43.5, 36.9, 36.7, 32.0, 31.0, 30.0 (2C), 14.6 ppm. UPLC-MS (ESI, m/z) Rt = 2.25 min –
494 (M+H)+; UPLC-MS purity (UV 215 nm): > 99.5%.
2-((5-(((3-cyano-6-methylpyridin-2-yl)oxy)methyl)-4-(3-phenylpropyl)-4H-1,2,4-triazol-3-
yl)thio)-N-cyclopentylacetamide
39.
2-((5-mercapto-4-(3-phenylpropyl)-4H-1,2,4-triazol-3-
yl)methoxy)-6-methylnicotinonitrile 66 (140 mg, 0.38 mmol), 2-chloro-N-cyclopentylacetamide 70l
(68 mg, 0.42 mmol), Cs2CO3 (137 mg, 0.42 mmol) and a catalytic amount of NaI in 5 mL of dry
MeCN were allowed to react for 4 h according to general procedure. The desired product
precipitated from water, then it was washed with water, Et2O and dried. Precipitate was purified by
flash chromatography, solvent system: DCM/EtOAc. Clean compound eluted with EtOAc 100%, to
give 39 (134 mg, 71% yield) as a white solid. 1H NMR (400 MHz, DMSO-d6) δ 8.19 (d, J = 7.8 Hz,
1H), 8.16 (br, 1H), 7.23 – 7.16 (m, 4H), 7.16 – 7.09 (m, 2H), 5.54 (s, 2H), 4.04 (t, 2H), 4.00 – 3.92
(m, 1H), 3.90 (s, 2H), 2.64 (t, J = 7.6 Hz, 2H), 2.10 – 1.99 (m, 2H), 1.81 – 1.67 (m, 2H), 1.65 – 1.53
13
(m, 2H), 1.53 – 1.41 (m, 2H), 1.41 – 1.27 (m, 2H) ppm. C NMR (101 MHz, DMSO) δ 165.9,
161.9, 161.3, 150.6, 144.1, 140.4, 128.2 (2C), 128.1 (2C), 128.0, 125.9, 117.4, 115.2, 92.4, 57.9,
50.7, 43.5, 36.7, 32.1 (2C), 32.0, 30.8, 24.3, 23.3 (2C) ppm. UPLC-MS (ESI, m/z) Rt = 2.24 min –
491 (M+H)+; UPLC-MS purity (UV 215 nm): > 99.5%.
4.1.3 Synthetic procedures to obtain hydrazides 43a-j
4.1.3.1 General procedure for synthesis of esters (73a-j). A mixture of appropriate amine 74a-i
(1 equiv), K2CO3 (1.05 equiv), and a catalytic amount of NaI in acetone was stirred at room
temperature. Benzyl, ethyl, or methyl bromoacetate 75, 76, 77 (1.1 equiv) was added dropwise to
the mixture, which was then refluxed for 2 h (Scheme 2, see also Supplementary data).
The hydrazide 43a was synthesized as previously described [24].
Benzyl 2-(5-fluoro-2-oxobenzo[d]thiazol-3(2H)-yl)acetate (73b). 5-fluorobenzo[d]thiazol-2(3H)-
one 74b (800 mg, 4.73 mmol), K2CO3 (687 mg, 4.96 mmol), benzyl bromoacetate 75 (810 ꢀL, 5.21
27

ACCEPTED MANUSCRIPT
mmol), and a catalytic amount of NaI in acetone (8 mL) were allowed to react according to the
general procedure.
The reaction was cooled down, water added, and aqueous phases extracted with EtOAc (3 X 10
mL). The collected organic phases were washed with brine, dried over MgSO4, filtered, and
evaporated in vacuo to give crude 73b (1.78 g) which was used in the next step without further
1
purification. H NMR (400 MHz, DMSO-d6) δ 7.78 – 7.71 (m, 1H), 7.46 – 7.40 (m, 1H), 7.40 –
7.35 (m, 5H), 7.15 – 7.07 (m, 1H), 5.21 (s, 2H), 4.94 (s, 2H) ppm. UPLC-MS (ESI, m/z) Rt = 2.37
min – 318 (M+H)+.
4.1.3.2 General procedure for synthesis of hydrazides (43a-j)
The appropriate ester 73a-j (1.0 equiv) was dissolved in 15 mL of EtOH and hydrazine hydrate
(23 equiv) was added. The reaction was left to stir at room temperature for 5-12 h. The reaction was
stopped, solvent evaporated under vacuum to obtain crude hydrazide, which was used in the next
step without further purification.
The hydrazide 43a was synthesized as previously described [24].
2-(5-fluoro-2-oxobenzo[d]thiazol-3(2H)-yl)acetohydrazide (43b). 73b (115 mg, 0. 63, 1 equiv)
was suspended in EtOH and hydrazine hydrate (150 ꢀL, 3.15 mmol, 4 equiv) was added. The
mixture was left to stir at room temperature for 5 h. Then reaction was stopped, and solvent
evaporated under vacuum to obtain crude 43b (150 mg), which was used in the next step without
further purification. UPLC-MS (ESI, m/z) Rt = 1.28 min – 242 (M+H)+.
4.1.4 General procedure for synthesis of isothiocyanates (44a-j)
1, 1-thiocarbonyldiimidazole 82 (1.2 equiv) was dissolved in dry DMF at 50 oC. To this mixture,
a solution of the appropriate amines 81a-j (1 equiv) and dry Et3N (1 equiv) in dry DMF was slowly
added dropwise. The mixture was stirred at 50 °C, and the progress monitored by TLC or
UPLC/MS. Then the reaction was cooled down, water was added, and the aqueous phase was
extracted with EtOAc (3 X 10 mL). Organic phases were washed with water and brine, dried over
28