New Insights into 4-Anilinoquinazolines as Inhibitors of Cardiac Troponin I-Interacting Kinase (TNNi3K).

We report the synthesis of several related 4-anilinoquinazolines as inhibitors of cardiac troponin I–interacting kinase (TNNi3K). These close structural analogs of 3-((6,7-dimethoxyquinazolin-4-yl)amino)-4-(dimethylamino)-N-methylbenzenesulfonamide (GSK114) provide new understanding of structure–activity relationships between the 4-anilinoquinazoline scaffold and TNNi3K inhibition. Through a small focused library of inhibitors, we observed that the N-methylbenzenesulfonamide was driving the potency in addition to the more traditional quinazoline hinge-binding motif. We also identified a compound devoid of TNNi3K kinase activity due to the addition of a methyl group in the hinge binding region. This compound could serve as a negative control in the study of TNNi3K biology. Small molecule crystal structures of several quinazolines have been solved, supporting observations made about overall conformation and TNNi3K inhibition.


Introduction
Kinases have been successfully targeted with 52 kinase inhibitors approved by the FDA to date. These kinase inhibitor drugs tend to be predominantly multi-targeted tyrosine kinase inhibitors for the treatment of cancer [1,2]. However, with over 500 kinases in the human genome, kinase inhibitors have therapeutic potential that extends beyond cancer [3], and there remains vast untapped potential within the kinome for treating a wide range of human diseases [4].To identify which kinases play key roles beyond oncology, highly potent and selective inhibitors are required [2].
One example of less-studied kinase is cardiac troponin I-interacting kinase (TNNi3K or CARK), a member of the wider tyrosine-like kinase (TLK) family. TNNi3K is selectively expressed in heart tissues and has been linked to cardiac hypertrophy dilated cardiomyopathy, pressure overload-induced heart failure, and ischemia/reperfusion injury in an in-vivo neonatal rat model [5][6][7]. A TNNi3K knockout mouse exhibited reduced ischemic injury suggesting that deletion of TNNi3K is cardioprotective [5][6][7]. These studies suggest that TNNi3K inhibition could serve as a unique strategy for addressing acute ischemic injury and heart failure [5][6][7].
Molecules 2020, 25, x; doi: FOR PEER REVIEW 2 of 13 One example of less-studied kinase is cardiac troponin I-interacting kinase (TNNi3K or CARK), a member of the wider tyrosine-like kinase (TLK) family. TNNi3K is selectively expressed in heart tissues and has been linked to cardiac hypertrophy dilated cardiomyopathy, pressure overloadinduced heart failure, and ischemia/reperfusion injury in an in-vivo neonatal rat model [5][6][7]. A TNNi3K knockout mouse exhibited reduced ischemic injury suggesting that deletion of TNNi3K is cardioprotective [5][6][7]. These studies suggest that TNNi3K inhibition could serve as a unique strategy for addressing acute ischemic injury and heart failure [5][6][7].

Synthesis of Aniline 13
To synthesize analogs of GSK114 (7) we first needed to access the previously reported aniline building block (Scheme 1) [8][9][10]. First 1-fluoro-2-nitrobenzene (9) was treated with neat chlorosulfonic acid at 0 °C and then refluxed for 2 h to afford the sulfonyl chloride derivative (10) in 69% yield (5 g scale). Next, 10 was then treated with methyl amine to afford the Nmethylbenzenesulfonamide product (11) in 81% yield (5 g scale) followed by a fluoride displacement using dimethylamine affording the nitro compound (12) in good yield (84%). The nitro group was reduced with hydrogen and Pd/C to furnish the aniline (13) in quantitative yield.  Scheme 1. Synthetic route to access non-commercial aniline 13.

Synthesis of Aniline 13
To synthesize analogs of GSK114 (7) we first needed to access the previously reported aniline building block (Scheme 1) [8][9][10]. First 1-fluoro-2-nitrobenzene (9) was treated with neat chlorosulfonic acid at 0 • C and then refluxed for 2 h to afford the sulfonyl chloride derivative (10) in 69% yield (5 g scale). Next, 10 was then treated with methyl amine to afford the N-methylbenzenesulfonamide product (11) in 81% yield (5 g scale) followed by a fluoride displacement using dimethylamine affording the nitro compound (12) in good yield (84%). The nitro group was reduced with hydrogen and Pd/C to furnish the aniline (13) in quantitative yield. One example of less-studied kinase is cardiac troponin I-interacting kinase (TNNi3K or CARK), a member of the wider tyrosine-like kinase (TLK) family. TNNi3K is selectively expressed in heart tissues and has been linked to cardiac hypertrophy dilated cardiomyopathy, pressure overloadinduced heart failure, and ischemia/reperfusion injury in an in-vivo neonatal rat model [5][6][7]. A TNNi3K knockout mouse exhibited reduced ischemic injury suggesting that deletion of TNNi3K is cardioprotective [5][6][7]. These studies suggest that TNNi3K inhibition could serve as a unique strategy for addressing acute ischemic injury and heart failure [5][6][7].

Synthesis of Aniline 13
To synthesize analogs of GSK114 (7) we first needed to access the previously reported aniline building block (Scheme 1) [8][9][10]. First 1-fluoro-2-nitrobenzene (9) was treated with neat chlorosulfonic acid at 0 °C and then refluxed for 2 h to afford the sulfonyl chloride derivative (10) in 69% yield (5 g scale). Next, 10 was then treated with methyl amine to afford the Nmethylbenzenesulfonamide product (11) in 81% yield (5 g scale) followed by a fluoride displacement using dimethylamine affording the nitro compound (12) in good yield (84%). The nitro group was reduced with hydrogen and Pd/C to furnish the aniline (13) in quantitative yield.  Scheme 1. Synthetic route to access non-commercial aniline 13. Scheme 1. Synthetic route to access non-commercial aniline 13.

Results of 4-Anilinoquinazolines 2, 7, 15-17 & 20 in TNNi3K
Binding Assay GSK114 (7) provided good baseline activity with an IC50 = 25 nM (Table 1) in an enzyme assay [10]. The incorporation of a methyl group in the hinge-binding region (15) led to complete loss of TNNi3K activity. The removal of the methoxy group in the 7-position (16) led to a decrease in potency by roughly 4-fold. Interestingly, removal of the other methoxy group at the 6-position (17) resulted in a 2-fold increase in potency. The removal of the dimethylamine substitution (2) on the aniline led to a nearly 5-fold loss of activity. However, removal of the N-methylbenzenesulfonamide (20) was more significant with an almost 250-fold drop in potency compared to GSK114 (7).   (Table 1) in an enzyme assay [10]. The incorporation of a methyl group in the hinge-binding region (15) led to complete loss of TNNi3K activity. The removal of the methoxy group in the 7-position (16) led to a decrease in potency by roughly 4-fold. Interestingly, removal of the other methoxy group at the 6-position (17) resulted in a 2-fold increase in potency. The removal of the dimethylamine substitution (2) on the aniline led to a nearly 5-fold loss of activity. However, removal of the N-methylbenzenesulfonamide (20) was more significant with an almost 250-fold drop in potency compared to GSK114 (7).

KinomeScan ® of GSK114 (7) and 15
GSK114 (7) and 15 were submitted for a KINOMEscan ® assay to explore the selectivity profile across >400 human kinases at a compound concentration of 1 µM. The screen (Figures 2 and 3) showed activity on only six kinases for compound 7 (with TNNi3K added from Lawhorn BG et al. [10]: MEK5, STK36, ZAK, GAK, PDFRB, BRAF and TNNi3K [10]) and two kinases (MEK5 and GAK) for 15. GSK114 (7) is a narrow-spectrum TNNi3K inhibitor but has some potent off-targets including GAK which has been previously reported for this chemotype [12][13][14][15]. GAK and MEK5 were also identified in the screening of the hinge-blocked analog (15), albeit significantly weaker activity than in the case of 7. If utilized together this set of compounds could be useful to investigate TNNi3K biology.

Modelling of Inhibitors in TNNi3K
The molecular modelling demonstrates that the N-methylbenzenesulfonamide interaction formed through a water-mediated bridge with the carboxylic acid of E509 and with the alcohol of T539 provides significant binding affinity in addition to the hinge contact ( Figure 4). Literature compounds 6 and 8 ( Figure 4A,B) show a binding pose mediated by the hinge-binding motif, but also by the N-methyl-benzenesulfonamide as a main interaction. The same binding interactions are also observed in compounds 17, 16, 2 in Figures 4C-E respectively, with the dimethylamine assisting in the conformational preference of the optimal inhibitor. Interestingly, despite the hinge binder being present in compound 20, the potency was significantly weaker, and this was supported by the modelling of 20 showing a loose, non-coordinated interaction of the aniline portion of the molecule in Figure 4F.

Modelling of Inhibitors in TNNi3K
The molecular modelling demonstrates that the N-methylbenzenesulfonamide interaction formed through a water-mediated bridge with the carboxylic acid of E509 and with the alcohol of T539 provides significant binding affinity in addition to the hinge contact ( Figure 4). Literature compounds 6 and 8 ( Figure 4A,B) show a binding pose mediated by the hinge-binding motif, but also by the N-methyl-benzenesulfonamide as a main interaction. The same binding interactions are also observed in compounds 17, 16, 2 in Figure 4C-E respectively, with the dimethylamine assisting in the conformational preference of the optimal inhibitor. Interestingly, despite the hinge binder being present in compound 20, the potency was significantly weaker, and this was supported by the modelling of 20 showing a loose, non-coordinated interaction of the aniline portion of the molecule in Figure 4F. Molecules 2020, 25, x; doi: FOR PEER REVIEW 6 of 13 One interesting observation of this focused library was the slight improvement in potency of 17 over 7 despite the very small difference with removal of just one methoxy group. In order to investigate this effect and rationalize the improved potency we simulated the bound water networks of 17 and 7 ( Figure 5) and found less disruption to the solvent network of 17 compared to 7. In the absence of direct interactions outside of simple Van der Waals contacts this provides a logical rationale for the increased potency observed with 17 [20]. One interesting observation of this focused library was the slight improvement in potency of 17 over 7 despite the very small difference with removal of just one methoxy group. In order to investigate this effect and rationalize the improved potency we simulated the bound water networks of 17 and 7 ( Figure 5) and found less disruption to the solvent network of 17 compared to 7. In the absence of direct interactions outside of simple Van der Waals contacts this provides a logical rationale for the increased potency observed with 17 [20].    17)). These hydrogen bonds link neighboring molecules in each stack with molecules in adjacent stacks via each chloride anion to form two-dimensional layers which stack via π-π interactions to form the structures. The structure of 20 comprises one-dimensional 'tapes' of 20H + hydrogen bonded to Cl − via the quinazoline N-H + and aniline N-H respectively (N-H + . . . Cl − . . . H-N (2.9732(9) Å, 3.1444(9) Å)) parallel to the b-axis. These tapes form two-dimensional, interleaved layers via π-π interactions parallel to the a-axis which close-pack to form the structure. The chloride anion is integral to the solid-state structures of each of the compounds (Figure 7).  17)). These hydrogen bonds link neighboring molecules in each stack with molecules in adjacent stacks via each chloride anion to form two-dimensional layers which stack via π-π interactions to form the structures. The structure of 20 comprises one-dimensional 'tapes' of 20H + hydrogen bonded to Clvia the quinazoline N-H + and aniline N-H respectively (N-H +… Cl -… H-N (2.9732(9) Å, 3.1444(9) Å)) parallel to the b-axis. These tapes form two-dimensional, interleaved layers via π-π interactions parallel to the a-axis which close-pack to form the structure.  (16,17) or b-axis (20). Hydrogen bonds are shown as dashed green lines.

Chemistry
All reactions were performed using flame-dried round-bottomed flasks or reaction vessels. Where appropriate, reactions were carried out under an inert atmosphere of nitrogen with dry solvents, unless otherwise stated. Yields refer to chromatographically and spectroscopically pure isolated yields. Reagents were purchased at the highest commercial quality and used without further purification. Reactions were monitored by thin-layer chromatography carried out on 0.25 mm E. Merck silica gel plates (60F-254) using ultraviolet light as the visualizing agent. NMR spectra were recorded on a Varian Inova 400 spectrometer (Varian, Palo Alto, CA, USA) and were calibrated using residual undeuterated solvent as an internal reference (CDCl3: 1 H NMR = 7.26, 13 (20). Hydrogen bonds are shown as dashed green lines.

Conclusions
We have demonstrated a robust way to access the 4-anilinoquinazoline scaffold in good yield supported by previous literature [12][13][14][15][16][17][18][19]. We report the synthesis of several related 4-anilinoquinazolines as inhibitors of cardiac troponin I-interacting kinase (TNNi3K). These close structural analogs of GSK114 (7) provide new understanding of structure-activity relationships between the 4-anilinoquinazoline scaffold and TNNi3K inhibition. Further, an interesting lesson in kinase design was found: having the hinge binder present does not always guarantee a potent compound, as in the case of 2 vs 20, where 20 without the N-methyl-benzenesulfonamide peripheral interaction is significantly weaker in TNNi3K inhibition. However, both interactions are required for a potent TNNi3K inhibitor. We also show that a simple hinge block with a methyl group (15) is enough to create a scaffold/negative control compound. We hope that the insights provide the medicinal chemist with an enhanced toolbox for kinase inhibitor design.