The assay performance was estimated using Z factors

The assay performance was estimated using Z′-factors (plotted in Fig. 2A) according to Zhang et al. calculated for each plate comparing positive (in the absence of an inhibitor) and negative controls (in the presence of an AdK inhibitor). The mean Z′-factor±SD was determined to be 0.7±0.1, indicative of a robust assay. The signal window (the signal-to-background (S/B) ratio) was calculated for each plate as a quotient of the mean minimal signal (negative control in the presence of an AdK inhibitor) divided by the mean maximal signal (positive control in the absence of an inhibitor). The mean value±SD for the signal window (ratio with inhibitor/without inhibitor) was calculated to be 0.23±0.12 across the plates (B). Altogether, these data demonstrated that the assay was reliable, reproducible, and well suited for HTS purposes representing a powerful tool for the discovery of novel adenosine kinase inhibitors. The AdK assay possesses several advantages, namely a simple experimental protocol, suitability for an HTS format, low sample consumption and low consumable costs. Next we applied the new assay to identify novel non-nucleoside adenosine kinase inhibitors by screening our proprietary focussed compound library biased towards purinergic targets. The compounds were initially tested at a concentration of 10μM (in some cases 1μM in case of limited solubility) for their potency to inhibit AdK. By screening only 189 selected compounds we successfully identified 12 hit compounds (showing an inhibition of >50% at 10μM) representing six new PIK-75 of non-nucleoside AdK inhibitors corresponding to a hit rate of 6% (Suppl. Fig. S2). The hit compounds 1–4 and 8 had been commercially obtained whereas compounds 5–7 and 9–12 had been synthesized in our laboratories in the framework of former purine receptor-related projects. The chemical structures and the IUPAC names of the obtained hit compounds are listed in Table 1. For these hit compounds full concentration–response curves using seven concentrations were determined (Fig. 3) and Ki values were calculated (summarized in Table 1). For each hit compound three independent experiments, each in triplicates, were performed. The most potent AdK inhibitor identified in the present screening campaign was compound 1 (10,11,12,13-tetrahydrobenzo[4,5]thieno[3,2-e]bis([1,2,4]triazolo)[4,3-a:4′,3′-c]pyrimidine-3,7-dithiol) with a Ki value of 184nM. Preliminary experiments indicated that compound 1 inhibits AdK in a reversible manner (Suppl. Fig. S3). The three structurally related derivatives 2, 3 and 4 lacking the second triazole ring were also found to be active in the AdK assay although they showed approx. 30-fold lower potency than compound 1. The determined Ki values for 2, 3 and 4 were 5.04μM, 5.44μM, and 5.64μM, respectively (Table 1, scaffold 1). Compound 5, a pyrazolo-triazolo-pyrimidine derivative (Table 1, scaffold 2) emerged as a further new chemical scaffold suitable for optimization. The two hit compounds 6 and 7 whose Ki values were calculated to be 8.75μM and 15.1μM, respectively, belong to the class of triazolo-quinazolines (Table 1, scaffold 3). 5-Methylbenzo[4,5]imidazo[1,2-c]thieno[3,2-e]pyrimidin-1-yl acetate (8) was also identified as an AdK inhibitor with moderate potency (Table 1, scaffold 4). Finally, the two pyrrolo-pyrimidine-diones 9 and 10 (Table 1, scaffold 5) as well as the two diaryl-oxazolo-pyrimidines 11 and 12 (Table 1, scaffold 6) were able to inhibit human AdK in the micromolar range and therefore represent two additional new scaffolds for the development of more potent AdK inhibitors. Our hit compounds show acceptable druglike properties (according to Lipinski’s rule of five). All were less than 500Da, carried less than 5 hydrogen bond donors and less than 10 hydrogen bond acceptors, and had calculated partition coefficient (LogP) values ranging from 0.196 to 5.346. The calculated parameters generated by the program StarDrop5 (Version 5.5 Build 222) are listed in Table 2.