T. Ekblad et al. / European Journal of Medicinal Chemistry 95 (2015) 546e551
547
Fig. 1. A: Phylogenetic tree of the human PARP-family ADP-ribosyltransferases. Enzymatic activities are indicated by symbols (black circles, poly-ADP-ribosylation; grey circles,
mono-ADP-ribosylation; rings, likely mono-ADP-ribosylation; crosses, putative inactive enzymes), B: Structural modifications made in the current program to target ARTD7, -8, and
-10, on primary hit compound 1 (cf. Table 1): positions I (amide), II (benzene ring), III (alkene), and IV (terminal carboxylic acid).
(Table 1) [20]. Here we describe a medicinal chemistry program
with the aim to develop potent inhibitors of ARTD7, -8, and -10.
investigate whether the amide position I (Fig. 1B) is equally
important for inhibition of ARTD7, -8 and -10. Moving the amide to
position 4 relative to the anilide (3) abolished inhibition, which
indicates that the amide in this position is crucial for inhibition of
both mono- and poly-ADP-ribosylation (Table 1). Functionalization
of the amide itself influenced the solubility dramatically, and
compounds 4e6 could not be analyzed in the enzymatic assay.
Subsequently, additional substituents were introduced on benzene
ring II (Fig. 1B, 7, 8, and 9) and also these modifications abolished
inhibition, indicating that the binding site accommodating the
benzene ring is restricted in size.
The role of the double bond in position III (Fig. 1B) was explored
by various modifications including saturation (10), addition of
methyl or methoxy groups (11, 12, 13 and 14), and ring formations
(15, 16, and 17). The chiral compounds 13, 15, and 26 were prepared
and evaluated as racemates. Most of these compounds retained
2. Results and discussion
Compounds 1 and 2 (Table 1) were discovered in a virtual screen
and verified as binders to ARTD7 and -8 by isothermal titration
calorimetry and x-ray crystallography [20]. We have now devel-
oped robust enzymatic assays for these two enzymes as well as
ARTD10 and could establish that both compounds indeed inhibit
the enzymatic activity of all three enzymes (Table 1). We found that
1 and 2 are more potent as ARTD10 inhibitors, with IC50 values of
1.3 mM and 10.6 mM respectively. Based on these promising results a
set of analogues of 1 and 2 (Table 1) was designed to explore the
structure-activity relationship (SAR) for inhibition of ARTD7, -8 and
-10 in relation to ARTD1. The compounds can readily be synthesized
from commercially available building blocks as outlined in Scheme
1. Aminobenzamides are reacted with carboxylic anhydrides or
carboxylic acids to give target compounds, and for a subset the
resulting carboxylic acid is further functionalized to esters or am-
ides. Fig. 1B illustrates the moieties explored within the SAR series
and Table 1 presents enzymatic inhibition data against ARTD7, -8
and -10 as well as ARTD1.
Previous data and crystal structures (e.g. Ref. [17]) indicated that
the benzamide in position I was likely to be crucial for enzyme
inhibition, and many known PARP inhibitors contains a benzamide
functionality that mimics the nicotinamide in NADþ [21]. The
amide forms a bifurcated hydrogen bond interaction to a conserved
backbone glycine in the active site of the enzymes. Comparison of
existing crystal structures indicated that the volume that harbors
the amide is similar between ARTD1 and the mARTDs, but there are
significant local differences in amino acid composition. To address
selectivity between ARTD1 and the mARTDs, we decided to
activity with profiles similar to those for 1 and 2, with mM potency
against ARTD10, ARTD7, and ARTD1. However, ARTD8 appears to be
more resistant to inhibition. Saturation was well tolerated (10) and
some alterations, such as the specific methylation in 12, seem
favorable for ARTD7 inhibition. Certain ring formations with
retained cis configuration (15, 16, and 17) were relatively well
tolerated. The carbon chain in position III (Fig. 1B) was then
extended (18) and, compared to the shorter chain in 10, this slightly
decreased the inhibition of ARTD7 and -10. The importance of the
carboxylic acid (position IV, Fig. 1B) was explored by modifying it
into alkylated amides (19, 20, 21 and 22), a methyl ester (e.g., 23) or
ketones (24, 25 and 26). To our surprise the methyl amide, 19, with
cis configuration could not be isolated using the standard synthetic
procedure. Instead we attempted a solid-phase synthesis strategy
according to the 9-fluorenylmethoxycarbonyl (Fmoc) protocol [22]
as outlined in Scheme 2. Using this method, 19 was successfully
synthesized. While this synthesis consists of more steps its overall