Isatin derivatives as a new class of aldose reductase inhibitors with antioxidant activity

In this work, isatin was employed as the scaffold to design aldose reductase inhibitors with antioxidant activity. Most of the isatin derivatives were proved to be excellent in the inhibition of aldose reductase (ALR2) with IC50 values at submicromolar level, and (E)-2-(5-(4-methoxystyryl)-2,3-dioxoindolin-1-yl) acetic acid (9g) was identified as the most effective with an IC50 value of 0.015 μM. Moreover, compounds 9a–h with styryl side chains at the C5 position of isatin showed potent antioxidant activity. Particularly, the phenolic compound 9h demonstrated similar antioxidant activity with the well-known antioxidant Trolox. Structure-activity relationship and molecular docking studies showed that the acetic acid group at N1 and C5 p-hydroxystyryl side chain were the key structures to increase the aldose reductase inhibitory activity and antioxidant activity.


Introduction
Diabetes mellitus (DM) is a complex metabolic disease characterized by chronic hyperglycemia caused by absolute or relative deficiency of insulin [1]. Patients with insulindependent and noninsulin-dependent DM, are vulnerable to chronic complications such as neuropathy, retinopathy, angiopathy, cataract, nephropathy [2]. Currently, the abnormally activation of glucose metabolism through polyol pathway and increased oxidative stress are thought to be important causes of diabetic complications [3].
As a member of the aldo-keto reductase superfamily, Aldose reductase (ALR2, EC 1.1.1.21) is the first and ratedetermining enzyme of the polyol pathway (Fig. 1). It catalyzes the NADPH-dependent reduction of various carbonyl compounds, including glucose [1]. Under normal circumstance, glucose is preferentially phosphorylated with ATP by hexokinase and then enters the glycolytic pathway. Due to the low substrate affinity of ALR2 for glucose, only a small amount of glucose enters the polyol pathway [4]. In the polyol pathway, glucose is reduced to sorbitol which is then oxidized to fructose by sorbitol dehydrogenase (SDH) [5,6]. Under hyperglycemic state, hexokinase is rapidly saturated and the polyol pathway is activated [7]. ALR2 converts the excess of intracellular glucose to sorbitol which is formed more rapidly than it is converted to fructose. The strong polarity of sorbitol makes it difficult to penetrate through membranes and subsequent removal from tissues by diffusion [2]. The abnormal accumulation of sorbitol may cause osmotic imbalance, membrane permeability changes and cell swelling and further develops into diabetic complications [8]. Studies have shown that aldose reductase inhibitors (ARIs) could prevent or slow the progression of DM-associated pathologies both in animal models and humans, which provides a promising strategy for the treatment of diabetic complications [9,10].
In addition to osmotic imbalance, increased oxidative stress is thought to be another main cause of diabetic complications [11]. There are multiple mechanisms that result in the induction of oxidative stress in DM [12]. Previous studies showed that abnormal activation of the polyol pathway was an important cause of oxidative stress [13][14][15].
Under hyperglycemic circumstance, above 30% of the glucose enters into aldose reductase-dependent polyol pathway, which consumes NADPH and consequently reduces GSH (intracellular antioxidant) level [13,14]. Moreover, oxidative stress occurs during the conversion of sorbitol into fructose by SDH while the cofactor NAD + is converted to NADH [13,15]. The sharp decrease of NADPH and NAD + causes changes of the redox potential in cells, which may lead to the increase of intracellular reactive oxygen species (ROS), ultimately resulting in cellular oxidative stress [16]. Additionally, fructose is metabolized into fructose-3-phosphate and 3-deoxyglucosone which are more potent non-enzymatic glycation agents [13]. These agents increase advanced glycation end products (AGEs) formation, eventually leading to ROS generation and oxidative stress [17]. Thus, ARIs can indirectly inhibit oxidative stress while constraining the intracellular accumulation of sorbitol [18]. Based on this, combining direct antioxidant activity with ALR2 inhibition in the design of new multifuctional ARIs maybe contribute to improve the efficacy of treatment.
During the past few decades, a great deal of ARIs have been reported, and some classical structure of them were shown in Fig. 2 [4]. According to functional groups, they were divided into three categories: sulfonyl derivatives, spiroimide derivatives, and carboxylic acid derivatives [4,5,[19][20][21]. As the most classic sulfonyl ARIs, ARI-809 exhibited outstanding inhibitory activity (IC 50 < 1 nM). However, the clinical manifestations of ARI-809 are still unknown [4,20]. Sorbinil and fidarestat were successful in the inhibition of ALR2 but gave rise to hepatotoxicity and hypersensitivity [4,19]. Compared with other inhibitors,  carboxylic acid derivatives serve as potent ARIs in which epalrestat is the only ARI in therapy [4,21,22]. It has been marketed in Japan, India, and China. However, some ARIs were limited in their activity, clinical efficacy, and some of them caused unwanted side effects [5]. The crucial reason for the side effects is the lack of selectivity relative to aldehyde reductase (ALR1, EC 1.1.1.2) which catalyzes the metabolism of some toxic aldehydes [23]. ALR1 and ALR2 belong to the same superfamily and have a high degree of homology in many aspects [24]. Ideal ARIs are expected to have little impact on ALR1 while inhibiting ALR2 effectively. Thus, the selectivity for ALR2 must be considered in the design of new ARIs.
Our research group have been committed to the development of new ARIs for many years [1,8,16,23]. Several groups of carboxylic acid ARIs based on structure of pyridothiadiazine, quinoxaline, benzothiadiazine, and pyridopyrazine were developed in recent years (Fig. 2) [16]. In these ARIs, the inhibition of aldose reductase and antioxidation were combined together [1]. Actually, oxidative stress and pathogenesis of the chronic diabetes complications are always interrelated [8]. Inhibiting the activity of ALR2 while reducing the progression of oxidative stress may be a new and effective therapeutic strategy [8]. Isatin (1H-indole-2,3-dione) is ubiquitous in nature, and its derivatives also possess a variety of pharmacological activity such as anticancer, antimalarial, and antioxidant, etc [25][26][27]. In addition, phenolic hydroxyl groups were widely used in drug design to enhance antioxidant activity [28]. Here, two series of isatin derivatives were synthesized to validate their inhibitory activity, selectivity, and antioxidant effect.

Results and discussion
As shown in Scheme 1, all required compounds were synthesized with 5-bromoisatin as the starting material. Since the Suzuki reaction on 5-bromoisatin is hard to achieve, the free carbonyl (ketone) group may be the key to solving the problem [29,30]. Thus 5-bromoisatin 1 was converted to the ketal 2 [31]. To afford the key intermediate esters 3, compound 2 was alkylated at the N1 position with methyl bromoacetate. Different phenyl groups were directly attached to the C5 position of the compound 3 by Suzuki-Miyaura reaction to provide compounds 4. Then, the desired carboxylic acids 5 were obtained by hydrolysis of 4 with lithium hydroxide. Compounds 5 were heated under reflux of hydrochloric acid to obtain isatin derivatives 6a-n possessing different substituents. To increase the space length of C5-side chain, styrene derivatives were coupled with 3 through a Heck cross-coupling reaction to produce compounds 7. The desired compounds 9 were prepared from 7 by the same method of hydrolysis and deprotection.
All of the prepared compounds were assessed for their potential to inhibit ALR2 extracted from rat lenses. In addition, the inhibitory activity against ALR1 isolated from rat kidneys was also tested to evaluate the selectivity of compounds for ALR2 inhibition.
All synthesized compounds were divided into two series according to the different side chains at C5 position of the isatin core. The first series of compounds contained phenyl residues (6a-n), while the second series contained styryl residues (9a-h). As shown in Table 1, most of the compounds significantly inhibited ALR2, and some of them (6c, 6j, 6m, and 9a-h) were even more potent than the positive control epalrestat. Compounds in the second series with C5-styryl side chain exhibited excellent inhibitory activity against ALR2 with IC 50 values varying from 0.015 to 0.075 μM, and they were stronger in efficacy than the first series with IC 50 values ranging from 0.064 to 0.591 μM. Besides, it was found that the vinyl segment between the phenyl and C5 positions of isatin had a positive effect on the improvement of inhibitory activity through the individual comparisons of the two series compounds. For example, 9a, 9b, 9c, and 9g were more active than their counterparts 6a, 6b, 6c, and 6j, respectively. Furthermore, all of the alkyl, methoxy, hydroxyl, and halogen substituents at the side of aromatic residue increased the inhibitory activity of the compounds (6b-n > 6a, 9b-h > 9a), which resulted from the changes of the electron cloud density on C5 aromatic rings. Therefore, the effect of compounds to inhibit ALR2 was significantly improved by introducing electron-withdrawing or electron-donating groups on the aromatic ring of the C5-side chain. In particular, methoxy group played an important role in ALR2 inhibition. Compounds (6j-m, 9g-h) possessing methoxy group demonstrated significant inhibitory activity and 9g was identified as the best inhibitor with an IC 50 value of 0.015 μM. By comparing compounds 6d with 6e and 6g with 6f, it can be concluded that compounds with para-substituents on the side chain had better enzyme inhibitory activity than those with meta-substituents.
All compounds were tested for their inhibitory ability against ALR1 at the concentration of 10 μM. As shown in Table 1, the inhibition percentage of all compounds was no more than 37.5%, which indicated slight effect on ALR1 and good selectivity for ALR2.
Oxidative stress plays an important role in the development of diabetic complications, and the antioxidant activity of compounds having a phenolic hydroxyl group on the side chain was investigated in the present study. Trolox was employed as a positive control. The results of antioxidant test were shown in Fig. 3.
The model reaction with the stable free radical of 1,1diphenyl-2-trinitrophenylhydrazine (DPPH) was applied to measure the free radical scavenging capacity of compounds. Compounds (9a, 9g, and 9h) containing vinyl spacer exhibited potential antioxidant activity with DPPH radical scavenging rates of 17.6, 7.9, and 94.4% at the concentration of 100 μM. Notably, 9h showed obvious DPPH radical scavenging activity similar to that of trolox. However, the scavenging effect of 6d and 6e on free radicals was insignificant. It showed that the vinyl spacer of the C5-side chain was essential for the DPPH radical scavenging, and it was consistent with the previous study [23]. It should be noted that 9h was more effective than 9a and 9g, indicating that  phenolic hydroxyl group played a key role in improving the DPPH radical scavenging activity. The position of the phenolic hydroxyl group had little effect on the antioxidant activity, which was deduced by the comparison of 6d and 6e.
To further evaluate the antioxidant activity of the tested compounds, the experiments of lipid peroxidation suppression were also implemented. As an indicator of lipid peroxidation, thiobarbituric acid reactive materials (TBARs) were produced by the reaction of thiobarbituric acid (TBA) with lipoperoxidation product malondialdehyde (MDA). The results of inhibiting lipid peroxidation test were consistent with the antioxidant results of the DPPH free radical scavenging test (Fig. 3b). Expectedly, compound 9h performed well-reduced production of TBARs as shown in Fig.  3b. The lipid peroxidation inhibition rate of 9h was close to that of trolox. As a consequence, compound 9h exhibited excellent activity in both enzyme inhibition and antioxidant tests, which may provide excellent inspiration for the development of lead structures for ARIs candidates.
According to the study of X-ray crystallography for ALR2, the ALR2 binding site was divided into three subregions: anion binding pocket, specificity pocket, and hydrophobic pocket [32]. Particularly, the specificity pocket is unique to ALR2 and is markedly different from the corresponding pocket of ALR1. Therefore, the specificity pocket is critical to increase the selectivity of compounds for ALR2 inhibition. Based on the above information, molecular docking studies were carried out by software Molegro Virtual Docker, version 5.0.
In order to study the interaction mode of compound with ALR2, the docking of compound 9h possessing the best activity both in the ALR2 inhibition and antioxidantion was performed with the human ALR2-NADP + -tolmetin complex (PDB code: 3S3G). As shown in Fig. 4a, compound 9h entered the cavity of ALR2 and bound tightly to the active site. It tucked the carboxyl group into the anion binding pocket and formed strong hydrogen bonds with Tyr48 (3.24 Å), His110  This indicated that the acetic acid group at N1 position of isatin played a key role in inhibiting ALR2. Besides, the free carbonyl (ketone) group of isatin core formed an additional hydrogen bond with the side chain of Tyr48 (3.09 Å). Therefore, the free carbonyl (ketone) group of isatin core may be contributed to the improvement of enzyme inhibitory activity. In addition, the 3-methoxy oxygen atom of C5-styryl ring was well embedded into the specificity pocket through two hydrogen bonds with Ser302 (3.13 Å, 3.13 Å), which explains its enhanced ALR2 inhibition. The C5-styryl ring was well placed into the specificity pocket, while the isatin scaffold matched very well with the hydrophobic pocket formed by the side chains of Trp219, Cys298, Trp79, Trp 20, and so on. Therefore, it is obvious that compound 9h was tightly fixed in the active site of ALR2 by interactions discussed above.
The docking behavior of 9h with ALR1-NADP + -fidarestat complex (PDB code: 3H4G) was also studied to illustrate its selectivity towards ALR2 over ALR1. The results showed that 9h did not match with ALR1. The carboxyl group at the N1 position of isatin totally "flowed out" of ALR1 active site, and the compound was almost separated from the surface of ALR1 (Fig. 4c). There was no significant interaction between 9h and the residues of the active site (Fig. 4d). Maybe that was the reason why compound 9h exhibited weak effect on ALR1.

Conclusions
In this work, two series of isatin carboxylic acid derivatives were designed as multifunctional ARIs and evaluated for their inhibitory activity and selectivity to ALR2 and antioxidant effects. Most compounds showed remarkable aldose reductase inhibition efficacy and marked selectivity, and 9g was considered as the most effective with an IC 50 value of 0.015 μM. Particularly, compound 9h containing 3-methoxy,4-hydroxyl group at the C5-styryl side chain exhibited the best activity in antioxidant test, which was comparable to that of trolox. The structure-activity relationship showed that the substitution group of C5-styryl side chain were conducive to improving the activity of the compound. In addition, phenolic hydroxyl group was installed in the side chain to reinforce the antioxidant property greatly. In conclusion, compound 9h containing C5 p-hydroxystyryl side chain demonstrated the best biological activity, which is expected to become a lead compound for multifunctional ARIs.

Experimental section
Chemistry TLC technique on silica gel Merck 60F254 was used to routinely check all reactions. Melting points of compounds were tested by the X-4 microscopic melting point apparatus. The hydrogen spectrum ( 1 H NMR: 400 MHz) and carbon spectrum ( 13 C NMR: 100 MHz) were obtained by Bruker Ascend 400 MHz nuclear magnetic resonance instrument. Deuterated reagents are CDCl 3 and DMSO-d 6 , with TMS as internal standard. An AGILENT LC/MS mass spectrometer was used for the determination of HRMS (ESI). The Hitachi D-2000 Elite HPLC system was used for analysis of sample purity. The HPLC conditions are as follows: room temperature; Inertsil ODS-2 250 × 10 mm, 5 mm column; mobile phase: CH 3 CN (0.1% TFA)/CH 3 OH = 75/25; flow rate is 1 mL/min, last for 8 min. The purity of all compounds used for biological assays was ≥95%.

Enzyme assays
Wistar rats (200-250 g) used in the experiment were supplied by SPF (Beijing) Biotechnology Co., Ltd. (certificate: 110229013482187, Beijing, China). According to the eported methods of Kinoshita J [33] and La Motta C [7], ALR1 and ALR2 were extracted from rat kidney and lens, respectively.
Compound solutions were prepared by dissolving the compounds in dimethyl sulfoxide (DMSO) and diluting to different concentrations. 5 μL of the compound solution was added to the above reaction mixture. To correct for the nonenzymatic oxidation of NADPH, the oxidation rate of NADPH in the presence of all reaction mixture components except the substrate was subtracted from each experimental rate. Firstly, the inhibitory effect of synthetic compounds was determined at a concentration of 100 μM (concentration in the final reaction system). Secondly, the inhibitory effect at other concentrations (between 100 μM and 10 nM) was determined. Most dose-response curves were generated using at least three concentrations of the compound with inhibitory activity between 20 and 80%, with three replicates at each concentration. Lastly, IC 50 values were calculated by linear regression analysis and regression equation (r 2 > 0.95).

DPPH assays
A method based on the DPPH scavenging of stable free radicals was used to investigate the antiradical activity of the tested compounds in a homogeneous system [34]. 100 µL of methanol solutions of various compounds (or Trolox) with different concentrations were added to the tube containing 1 mL of DPPH methanol solution (0.025 mg/ mL) and 1.9 mL of methanol solution, so that the final concentrations of the tested compounds were 100, 10 and 5 µM, respectively. The composition of the control group was DPPH methanol solution (1 mL, 0.025 mg/mL) and methanol (2 mL). The methanol solution (0.1 mL) of the testing compound and methanol (2.9 mL) constituted a blank group. The above reaction system were vortexed thoroughly and placed at room temperature for 120 min. The absorbance was measured at λ 517 nm by the Shimadzu UV-1800 spectrophotometer. The percentage of DPPH radical scavenging was calculated by the following equation. Experiments were performed in triplicate.

Lipid peroxidation inhibition
The brain homogenate was prepared by crushing freshly isolated rat brains with ice-cold normal saline. Then the homogenate was centrifuged at a speed of 3000 rpm for 10 min to obtain the supernatant for biochemical analysis. The protein concentration in the supernatant was determined by a total protein quantification kit. The reaction mixture containing the title compounds (100 μM), FeCl 3 (0.02 μM), ascorbic acid (0.1 μM) and brain homogenate supernatant was incubated at 37°C for 30 min. After that, TBA was mixed with the reaction mixture and then centrifuged to obtain supernatant. The concentration of MDA in the supernatant was determined by a commercially available TBA-based kit. All the subsequent steps were followed according to the instruction of manufacturer [35,36]. The supernatant and solution in the MDA kit were mixed in a centrifuge tube and boiled at 95°C for 40 min. Then, the centrifuge tubes were cooled and centrifuged (4000 rpm, 10 min). The red product produced by the reaction was measured with a spectrophotometer at λ 532 nm. Experiments were performed in triplicate.

Docking studies
Docking was performed by Molegro Virtual Docker (version 5.0) with the same protocol as described before [23].