Development of 2-(4-pyridyl)-benzimidazoles as PKN2 chemical tools to probe cancer

Graphical abstract

T PKNs have a fairly conserved primary sequence and they share the same architecture. The catalytic domain of PKN2 has 87% percent identity with PKN1; 70% with PKN3; and 50% with PKC kinases, while the N-termini regions are less conserved, sharing only 48% and 40% between PKN1/2 and PKN2/3, respectively. 7,8 PKNs have been linked to various cellular roles, including cytoskeleton regulation, 9 transport, 10 cell adhesion, 11 nutrient signalling, 12 and cell cycle, 13 as well as being a target of interest in colon, 14 breast, 15 renal, 16 head, 17 neck, 17 and prostate cancers. 18 They are also reportedly involved in inflammation 19,20 and heart failure. 21 So far, there is one Xray crystal structure of PKN2 publicly available in the Protein Data Bank (PDB ID: 4CRS) (Fig. 2).
These previous studies have elucidated functions for PKN2 using molecular and cell biology techniques, and the conclusions would be greatly supported by validation through the use of small molecule inhibitors, especially to evaluate PKN2′s potential as a cancer drug target. Potent inhibitors are known for several AGC kinase family members, including ROCK 22-25 and PKC, 26 but currently there are no sufficiently selective inhibitors for PKN2. 12 This work describes an initial effort to develop such compounds based around a benzimidazole core. Compound 5 was previously developed as a PARP inhibitor 27-29 but exhibited higher potency towards PKN2 than its desired target. Benzimidazoles are N-containing heterocycles that are prevalent in medicinal chemistry. 30 The compound was found as part of a screen of the Abbott chemical library 31 via the ChEMBL database when searching for PKN2 inhibitors. It had a reported K i of 0.040 μM against PKN2 while only inhibiting two out of 137 other kinases (PKN1 and CLK4) with potencies lower than 0.100 μM. 31 This was deemed a good starting point for repurposing the compound as a PKN2 inhibitor. We report the synthesis of that compound and subsequent SAR studies to determine its viability as a chemical tool for establishing the potential of PKN2 as a therapeutic target.
Compound 5 was successfully synthesised via a four step synthesis (Scheme 1). 2-Amino-3-nitro-benzoic acid (1) was treated with ammonia and CDI-coupling conditions 32 to form amide 2. The 3-nitro group was reduced to aniline 3 with sodium dithionate, 33 followed by the coupling of isonicotinic acid to the 3-position aniline to form amide 4, 34 which was then heated in acetic acid to form benzimidazole 5. 35 The scope of this chemistry enabled the synthesis of 14 analogues using commercially available nitroanilines and di-anilines. Additional alkylation conditions allowed the capping of the benzimidazole N-H 36 (6) and alternative amide coupling conditions were used for preparing compound 11 37 and the penultimate amide intermediate used to make compound 19. 38 The potencies and selectivities of these compounds were tested using a TR-FRET binding-displacement assay in which the IC 50 values were measured ( Table 1). Calculation of K i values using the Cheng-Prusoff equation and the K D of the tracer (previously determined) allowed the affinity of the inhibitors for PKN2 and PKN1 to be compared (Table 1).
Compound 5 was validated as a PKN2 inhibitor (K i = 0.032 μM) with 17-fold selectivity over PKN1 (K i = 0.500 μM) which was not previously included in the Abbott library screen used in the Metz et al. study. 31 The benzimidazole NeH was capped using chemistry described by Tsukamoto et al. 36 While the alkylation conditions given were said to be applicable to methylation of the benzimidazole using the corresponding methyl halide, this proved unsuccessful; a dimethylated product formed instead, thought to be due to the susceptibility for the 4′pyridyl to also alkylate after the benzimidazole NeH. Repeating the specific reaction conditions used by the authors incorporated a methyl acetate ester at the 1-position (6) which led to loss of binding to PKN2.  Moving the 4′-pyridyl nitrogen in 7 and 8 resulted in loss of activity, as did introducing an electron-donating methoxy group at the 3′-position (9). This suggests the 4′-pyridyl ring acts as the hinge binder. Attempts to make the 2′-pyridyl and 4′-pyrimidine analogues were unsuccessful (Scheme 2).
Capping the amide with one (10) or two (11) methyl groups led to increasing loss of activity respectively. Potency was lost when the amide was moved to the 5-position of the benzimidazole ring (12), Removing the amide completely (13) or exchanging the 4-or 5-position for another functional group (14-18) also led to loss of activity.
Introduction of a bromine at the 6-position (19) was hoped to provide a useful handle for incorporating various alkyl/aryl groups at that position using Suzuki coupling chemistry. 39,40 This reaction was attempted at multiple stages of the synthetic route but was unsuccessful. Compound 19 was active against PKN2 but was nearly three times less potent than compound 5. Despite this reduction in potency, compound 19 is 26-fold selective over PKN1.
The SAR exploration around 5 confirms that the primary amide at the 4-position, 4'-pyridyl and free NeH at the 1-position are necessary for the compound's activity against PKN2. Subsequent analogues prepared for this series did not improve potency for the target within the PKN family but did result in a slight improvement in selectivity over PKN1 in compound 19.
Chemical tools are needed to facilitate the exploration of lesser understood kinases such as PKN2 for its roles in healthy and cancerous cells. Benzimidazole 5 was validated as an inhibitor of PKN2 with IC 50 0.064 μM and with ca. 17-fold selectivity over PKN1 with reported high selectivity across the wider kinome 31 . Our efforts to develop a new compound to inhibit PKN2 resulted in compound 19 which was 26-fold selective for PKN2 over PKN1 despite having a near three-fold reduction in potency compared to compound 5.

Acknowledgements
This work was supported by a Continuing Excellence Fund from the Genome Damage and Stability Centre, University of Sussex. Thanks also to additional funding from the Wellcome Trust for initial assay experiments.
This work was also supported by the Brazilian agencies FAPESP (Fundação de Amparo à Pesquisa do Estado de São Paulo) (2013/ To the best of our knowledge there are no competing interests with involved parties.