Synthesis and multifaceted pharmacological activity of novel quinazoline NHE-1 inhibitors

The Na+/H+ exchanger isoform 1 (NHE-1) attracts ongoing attention as a validated drug target for the management of cardiovascular and ocular diseases owing to cytoprotective, anti-ischemic and anti-inflammatory properties of NHE-1 inhibitors. Herein we report novel NHE-1 inhibitors realized via functionalization of N1-alkyl quinazoline-2,4(1H,3H)-dione and quinazoline-4(3H)-one with N-acylguanidine or 3-acyl(5-amino-1,2,4-triazole) side chain. Lead compounds show activity in a nanomolar range. Their pharmacophoric features were elucidated with neural network modeling. Several compounds combine NHE-1 inhibition with antiplatelet activity. Compound 6b reduces intraocular pressure in rats and effectively inhibits the formation of glycated proteins. Compounds 3e and 3i inhibit pro-inflammatory activation of murine macrophages, LPS-induced interleukin-6 secretion and also exhibit antidepressant activity similar to amiloride. Hence, novel compounds represent an interesting starting point for the development of agents against cardiovascular diseases, thrombotic events, excessive inflammation, long-term diabetic complications and glaucoma.


Results
Synthesis. Target compounds were realized via the synthetic route shown in Fig. 2. Starting N 1 -substituted quinazoline-2,4 (1H,3H)-diones 1 were readily alkylated with esters of bromoacetic and D,L-2-bromopropionic acid at room temperature in an anhydrous DMF medium in the presence of the excessive amount of potassium carbonate. Corresponding esters 2a-f were obtained in 67-81% yield. The next step involved treatment of esters 2a-f with guanidine generated in situ from guanidine hydrochloride and potassium hydroxide in boiling 95% ethanol, which leads to rapid cleavage of the ester bond and formation of N-acyl derivatives of guanidine 3a-d in 57-84% yield.
When aminoguanidine was used as a nucleophilic reagent, which was similarly obtained in situ from aminoguanidine carbonate and potassium hydroxide in boiling 95% ethanol, the reaction is accompanied by cyclization to form 5-amino-1,2,4-triazole and leads to quinazoline-2,4(1H,3H)-dione derivatives 3e-i with a yield of 60-81%. Ester derivatives of quinazoline-4(3H)-one 5a and 5b analogously react with aminoguanidine to form 5-amino-1,2,4-triazole derivatives 6a and 6b. NHE-1 inhibition. Firstly, newly obtained quinazoline-2,4(1H,3H)-dione and quinazoline-4(3H)-one derivatives were assayed for NHE-1 inhibition. Compounds 3a-d comprising acyclic guanidine fragments showed moderate activity (Table 1). Structure-activity analysis suggests that bulky allyl and benzyl N 1 -substituent favors methyl group as R 3 , while small methyl at N 1 can only tolerate unbranched side chain (i.e., R 3 = H). Bromine at C 5 did not influence the activity. Derivatives 3e-I, 6a and 6b that contain 5-amino-1,2,4-triazole in N 3 side chain generally showed better efficacy (Table 2). Previously made SAR observation is also evident here, since methyl R 3 in the linker region decreases the NHE-1 inhibitory activity for N 1 -methyl-substituted compounds, but improves it for N 1 -allyl counterparts. Structural similarity of the most active compounds 3e and 6a (R 2 = R 3 = H) suggests that R 1 and C 2 carbonyl are dispensable for NHE-1 inhibition.   The combination of these binding points results in a pharmacophore that determines a high level of NHE-1 inhibitory (Fig. 3). Incorporation of the constructed pharmacophore into the structures of the two most active compounds and zoniporide are shown in Table 4. The structures of compound 6a and zoniporide contain complete pharmacophores of 17 and 24 entries of four types of QL-descriptors of a high level of NHE-1-inhibitory activity, respectively. The structure of compound 3e includes 27 entries of three types of QL-descriptors of this pharmacophore, and all of the found pharmacophore fragments occur several times. Despite that zoniporide was excluded from the training set, the pharmacophore model obtained correctly reproduce its pharmacophore features. Therefore, we assume that the pharmacophore is valid and can be employed to design novel potent NHE-1inhibitors. Anti-inflammatory activity. As a next step of the study target compounds were pharmacologically profiled for NHE-1 associated activities. A study of the anti-inflammatory activity was carried out on primary peritoneal macrophages of C57BL/6J mice (Fig. 4). Three compounds (3d, 3e, 6b) were found to statistically significantly inhibit the synthesis of nitric oxide at a concentration of 100 μM in the absence of a statistically significant effect on cell viability. The secretion of interleukin 6 was suppressed by four compounds -3b, 3c, 3e, 3i. After a concentration-response validation study (Fig. 5), the lead compound 3i was identified, which inhibits Table 2. NHE-1 inhibition by novel quinazoline-2,4(1H,3H)-dione and quinazoline-4(3H)-one derivatives with cyclic guanidine moiety. Statistical significance: *p < 0,05 vs. negative control; # p < 0,05 vs. Zoniporide (1-way ANOVA). a Not tested. b Not determined.  Table 3. Neural networks were obtained after iterative modeling. MLP-multilayer perceptron; k-m-1-the number of input, hidden and output neurons; BFGSN-an algorithm for finding the minimum of the error function; Exp, Tanh, Ident-activation functions of the hidden and output layers of neurons, exponential, hyperbolic tangent, identical, respectively.   Antiplatelet activity. In addition, it was shown that the majority of them exhibits pronounced antiplatelet activity (Fig. 3). Unfortunately, the most active NHE-1 inhibitors 3e and 6a were insoluble under assay conditions. Four compounds (6b, 3a, 3f and 3h) exceeded the activity of the reference drug, acetylsalicylic acid. Noteworthy, cyclic guanidine analogs again proved to be more active than compounds with acyclic guanidine side chains. Influence of N 1 substituent on antiplatelet activity appears to be as follows: no substituent (6b) or methyl (3a, 3f) > allyl (3h) > benzyl (3c, 3i). Neither bromine at C 6 , nor methyl in the linker region (R 3 ) had a definite effect on the activity.
IOP-lowering activity. Effect on intraocular pressure (IOP) was studied on albino outbred rats. Two compounds (3d, 3e) exhibit statistically significant activity, though the effect size is negligible for 3e, while 3d is comparable to zoniporide and superior to timolol. The complex anatomical structure of the eye presents a barrier for local drug delivery, which might explain a lack of correlation between NHE-1 inhibitory (and other cellbased assays) and IOP-reducing activities. Interestingly, the most active compound 3d is also the most lipophilic one among the identified NHE-1 inhibitors. This is in line with several studies reporting that drug lipophilicity expressed as logD enhances both corneal and conjunctival permeability 26,27 . Antiglycating activity. Additionally, we have found that several compounds can effectively prevent the formation of advanced glycation products from bovine serum albumin and glucose. The most active compounds 6b showed better potency than reference antiglycating drug aminoguanidine (IC 50 s 506.3 and 4521.2 µM, respectively). Since investigated compounds share similar nucleophilic centers and chelating abilities, which usually mediate antiglycating properties 28,29 , we were unable to rationalize SAR in this case. Presumably, higher activity is associated with the accessibility of the primary amine group of hydrazine (amiloride). Acyclic guanidine group instead of hydrazine is associated with lower activity 30 .

Antidepressant activity.
Suspension by the tail test was used to assess the antidepressant activity of lead compounds. It was found that compounds 3e and 3i exhibit antidepressant activity similar to amiloride 7 , although weaker than comparison drugs imipramine and amitriptyline.

Discussion
NHE-1 inhibitors are experimentally and clinically validated pharmacological agents especially as potential cardioprotective therapies, agents against glaucoma and other disorders associated with ischemia and reperfusion, cell proliferative disorders and diabetes 31,32 . Modulation of sodium-hydrogen exchange prevents intracellular Ca 2+ mobilization ameliorating ischemia-reperfusion cellular damage and platelet activation, thus NHE-1 inhibitors have cardioprotective 33 , neuroprotective 34 and anti-arrhythmic properties 35 . Recently, guanidinecarbonyl 36 and (imidazol4-yl-carbonyl)guanidine 37 derivatives have been described as potent NHE-1 inhibitors, however, their pharmacological evaluation was limited to cardioprotective effect. Herein we report a novel series of guanidine-modified quinazoline s as NHE-1 inhibitors that also possess anti-inflammatory and antiglycating properties, both of them being important additional mechanisms of action in conditions of systemic inflammation and diabetes mellitus, which are in turn associated with increased cardiovascular risk and retinopathy. Compounds 3a, 3d, 3h have activity comparable to the reference drug rimeporide, while 3e and 6a are more potent and comparable to zoniporide. NHE-1 inhibition was confirmed in follow-up cellular and animal studies, although lack of apparent correlation with NHE-1 inhibition is likely due to cellular permeability and uptake factors.  6a Zoniporide Total -17 Total -24 Scientific Reports | (2021) 11:24380 | https://doi.org/10.1038/s41598-021-03722-w www.nature.com/scientificreports/ Identified active compounds prevent ADP-induced platelet aggregation, LPS-induced macrophage activation, reduce intraocular pressure and also demonstrated some antidepressant potential, probably via amelioration of microglia activation 5 . We have identified NHE-1 inhibitors 3e and 3i to prevent LPS-induced macrophage polarization and IL-6 secretion in micromolar range with 4-6 selectivity over cytotoxicity. Along with this, 3e and 3i possess antidepressant activity comparable with amiloride. ADP-induced platelet aggregation was most effectively inhibited by compounds 3d, 3h and 6b, which exceeded acetylsalicylic acid. NHE-1 inhibitor 3d was also found to markedly reduce intraocular pressure. In conclusion, we posit that the identified pharmacophore may serve as a viable starting point for development of novel NHE-1 inhibitors endowed with valuable pharmacological activities. Further efforts will be undertaken to optimize the structure and potency of quinazolinederived NHE-1 inhibitors.

NHE-1 inhibition assay.
Evaluation of NHE-1 inhibition was carried out on rabbit platelets by the known method 39,40 . The experiments were carried out on 15 male rabbits weighing 3.0-4.0 kg. Platelet-rich plasma (PRP) was obtained by centrifuging blood with 3.8% sodium citrate (1:10) at 1000 rpm for 12 min (Multi centrifuge CM 6M, Latvia). Platelet shapeshifting due to acidification was followed with a laser aggregometer BIOLA-220 LA (Russia). Test compounds (10 μl, 10 nM final concentration) were added to 200 μl of PRP 5 min in a cuvette before the addition of sodium propionate solution, incubated with constant stirring using a magnetic stirrer (800 rpm, 37 °C). Pharmacophore modeling. Structures of 13 target compound were characterized with a matrix of QLdescriptors of the 2nd rank of the 5th type using IT Microcosm system 41 . QL-descriptor matrix and experimental values of NHE-1 inhibitory activity activity for these substances served as an initial training set. We used a two-layer perceptron with a bottleneck MLP k-m-1 for neural network modeling of regression dependence, where the number of input neurons k >> m is the number of hidden neurons; calculations were performed in the Statistica program 42 . Iterative training of networks was performed with the division of the initial dataset into training and test sets in a ratio of 60/40% with an automatic selection of neural networks with high values of the correlation coefficients. Training of each of the 500 neural networks involved random selection of training and test sets to minimize possible bias. For the best performing neural network, sensitivity analysis (Sens) of input neurons was performed, low-sensitivity neurons were removed, and iterative neural network modeling was performed. In the best neural network, the most sensitive input neurons were identified. By superposition of the found significant QL-descriptors, a pharmacophore was formed that defines a high level of NHE-1 inhibitory activity of the tested compounds. Zoniporide structure was excluded from training sets but was used to validate the model.
Platelet aggregation assay. Functional activity of platelets was determined on a two-channel laser analyzer of platelet aggregation "BIOLA-220 LA" (Russia) as described previously 43 . The experiments were carried out on 6 male rabbits weighing 3.5-4 kg. To prepare platelet-rich plasma (PRP) venous blood was taken from the ear marginal vein of a rabbit, stabilized with a 3.8% sodium citrate solution in a ratio of 9:1 and centrifuged for 10 min at 1500 rpm. PRP (300 μL) and a solution of the test compound at a concentration of 100 μM were sequentially introduced into the cell of the aggregometer. The samples were incubated in thermostated cells of the aggregometer at 37 °C for 5 min. To induce aggregation adenosine-5-diphosphoric acid (ADP, Sigma, USA) at a final concentration of 5 μM, was added to the cuvette. Acetylsalicylic acid (Shandong Xinhua Pharmaceutical Co., Ltd., China) was used as a reference drug. Experiments were run in 5 independent replicates. Assay of interleukin-6 (IL-6). Cell supernatant was collected and centrifuged at 1000 g for 20 min in a 2-16PK Sigma centrifuge (Germany). The concentration of IL-6 was determined by ELISA using a commercial kit (Cloud-clone ELISA kit) with a microplate reader Infinite M200 PRO (Tecan, Austria).
Cytotoxicity study. The activity of lactate dehydrogenase (LDH) in a cell culture medium served as a marker of membrane permeability and cell death. Aliquotes of supernatants were taken after 24 h of inoculation with test compounds, mixed with 250 μl of 0.194 nM NADH solution in 54 mM phosphate buffered saline (pH 7.5). Then, 25 μl of a 6.48 mM pyruvate solution was added to the mixture. The optical density was followed at a wavelength of 340 nm for 20 min. Conversion of optical density into cell viability was carried out according to a standard curve (DMSO-treated cells as 100% and 0.01% Triton X-100-treated cells as 0% viable cells).

Intraocular pressure studies.
The experiments were carried out on 75 adult outbred rats of both sexes, which were kept in standard cages at a temperature of 25 °C and a standard light regime. Animals were randomly assigned to the control and experimental groups (n = 5). Before the start of the experiment, the rats had free access to food and water. Intraocular pressure (IOP) was measured with a TonoVet device (Finland) 44 which measures IOP with a short touch of a small disposable tip in the center of the cornea, no corneal anesthesia is required. The study was carried out according to the published method 45 . One drop (50 μl) of a test 0.4% solution of the compound was instilled into the right eye, and deionized water was added to the left eye. IOP was measured in both eyes. The left eye, in turn, serves to assess the possible systemic exposure of the test compounds. Drugs with NHE-1-inhibitory activity were instilled in concentrations suggested in therapeutic practice -0.2% zoniporide, 0.4% amiloride. IOP was measured at five time points (0, 1, 2, 3, and 4 hrs), where 0 hour is the baseline value. IOP-lowering activity was assessed as the maximum decrease in IOP from the initial values. At 9:00 AM, baseline IOP was measured in animals of all groups. Experiments were run in 5 independent replicates.
Study of antiglycating activity. The glycation reaction was modeled using 1 mg/ml bovine serum albumin (Chemmed, Russia) and 0.36 M glucose (Vekton, Russia) in 50 mM phosphate buffer solution (PBS, pH 7.4) at 60 °C. After 24 hrs, albumin was precipitated using trichloroacetic acid (10% final concentration) and centrifugation (15 000 rpm, 4 min). The supernatant was removed by aspiration, and the residue was dissolved in 50 mM PBS (pH 10.5). Aliquots of 300 µl were transferred to a black flat-bottom 96-well microplate. AGE fluorescence was registered using Infinite M200 Pro (Tecan, Austria) microplate reader at excitation/emission wavelengths of 370/440 nm. Signals were normalized using blank samples containing BSA and the test compound in the appropriate concentration without the glucose. When normalized, the activity is expressed as the AGE fluorescence coefficient, determined by the formula: where A and B correspond to the absolute fluorescence values of the glucose-containing and mean glucose-free sample. Activity percentage was calculated using the formula: where Flu(Sample) and Flu(Control) correspond to the normalized fluorescence coefficients of experimental and control samples. Experiments were run in 5 independent replicates.
Tail suspension test. The experiment was carried out on 36 male ICR mice weighing 22-25 g, divided into groups of 6 animals each under room lighting (300 lux). Tail suspension test 46 was performed using the Panlab LE808 apparatus, suspended by the tail with a piece of adhesive tape (painless method), instinctively tries to free itself from an unpleasant situation, after unsuccessful attempts to escape, the animal begins to demonstrate the behavior of despair -immobilization. At the same time, it is believed that the severity of despair, determined by immobility, directly depends on depressive disorders in the subjects, and is significantly reduced when taking antidepressants. In experiments, test substances were administered intragastrically 30 min before the start of the test. Experimental groups received test compounds in an equimolar dose to the comparison drug amiloride. Animals of the remaining groups were treated with 2.6 mg/kg amiloride, 10 mg/kg amitriptyline, or 8 mg/kg imipramine. Distilled water was administered to the control group. The time of immobilization of the animals was recorded.