Hypoglycemic potential of cyclic guanidine derivatives

Introduction

Type 2 diabetes mellitus (T2DM) is a widely recognized medical and social concern, which is due to the high prevalence, growing incidence, chronic nature, high disability and mortality rates of patients due to development of late vascular complications (micro- and macroangiopathies), as well as a need for specialized care system [1]. No less alarming is the fact that T2DM is ranked third among the primary death causes after cardiovascular diseases and cancer [2], [3], [4]. According to statistics, every 6–7 diabetes patients out of 10 die from the macroangiopathic complications. Accordingly, the socio-economic impact associated with the development of severe disabling complications is growing [5].

The current armamentarium against T2DM includes the following groups of drugs with a high level of clinical evidence: insulin secretagogues (sulfonylureas glibenclamide, glipizide, gliclazide, gliquidone, glimepiride); postprandial insulin secretion regulators (repaglinide and nateglinide); biguanides (metformin); insulin sensitizers (rosiglitazone, pioglitazone); incretin mimetics (vildagliptin, sitagliptin, saxagliptin). Nevertheless, there is an evident urge for more potent pharmacotherapies with a novel mechanisms of action [6], [7], [8], [9].

Cyclic guanidines as a source of antidiabetic agents

Presence of a guanidine moiety in well-known alkyl and aralkyl biguanides (metformin, buformin, phenformin) has made it an indispensable anchor for the development of new antidiabetic agents. One of the first extensive studies in the field was performed by Ishikawa, Koasasayama and coauthors. They described synthesis and biological evaluation of a broad chemical class which is defined as a “cyclic guanidines” (Fig. 1), particularly, monocyclic imidazolines (1) [10], bicyclic imidazo[1,2-a]pyrimidines (2) [11], and tricyclic imidazo[2,1-b]quinazolines (3) [12]. Structure-activity studies revealed that guanidine fragment is a prerequisite for hypoglycemic activity, while bulky benzyl or benzhydryl substituents maximize insulinogenic properties. Also, a number of compounds were found to inhibit platelet aggregation.

Fig. 1:

Cyclic guanidines possessing hypoglycemic properties.

Through serendipity, it was observed that antihelmintic drug mebendazole (4) improves glycemic control in diabetic patients through stimulation of glucose-induced insulin secretion [13], [14]. This finding influenced further research on a potential of 2-aminobenzimidazole scaffold [6], which is also considered as a cyclic guanidine.

Later, Anisimova et al. [15] reported the 2,3-dihydro-imidazo[1,2-a]benzimidazoles (5) as a hypoglycemic agents. Introduction of a methyl, ethyl or butyl as R substituent rendered the maximal hypoglycemic effect 4 h after administration. In case of a propyl or piperidinoethyl substituents the maximum effect was developed after 2 h. All 2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives 5 exceeded chlorpropamide (4-chloro-N-(propylcarbamoyl)benzenesulfonamide), but were less active than buformin (1-butylbiguanide).

As a development of this work, 1-aminoethyl-2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives (6) were synthesized and evaluated to explore the influence of the N¹ versus N⁹ substitution [16]. It was established that dihydrochlorides 6 bearing pyrrolidine or piperidine moieties as NR¹R² caused statistically significant reduction of blood glucose levels exceeding that of the reference drugs chlorpropamide and buformin. Introduction of morpholine, cyclohexylamine or phenylamine residues is associated with a slightly lower activity comparable with chlorpropamide.

Later, several works were performed to confirm hypoglycemic potential of benzimidazoles. Among 14 synthesized 3-aroyl and 3-hetaroyl-imidazo[1,2-a]benzimidazoles (7) 6 compounds were found to lower blood glucose [17]. Noteworthy, derivatives comprising the N⁹-(diethylamino)ethyl moiety are more potent than (piperidino)ethyl counterparts.

A series of aroylmethyl-substituted benzo[d]imidazo[1,2-a]imidazoles and related derivatives 8 were synthesized and found to be hypoglycemic as well. Structure-activity relationship study revealed that 4-hydroxyphenyl as R produces marked and sustainable hypoglycemia greatly exceeding the reference drug chlorpropamide [18].

A variety of heterocyclic compounds with a broad spectrum of pharmacological actions including hypoglycemic properties were obtained by introduction of an acetamide fragment [19]. It was shown that certain 3-(N,N-disubstituted)acetamido-N¹-substituted 2-amidobenzimidazolium chlorides produce marked and prolonged hypoglycemic effect. The highest hypoglycemic activity within 6-h period was associated with lipophilic NR²R³ moiety, e.g. alkyl substituted phenyl (4-isopropyl, 2,6-dimethyl), along with alkyl present in R¹ position. These substances (9 and 10) surpass the potency of the reference drugs glibenclamide and metformin in experimental animals. Replacement of the alkyl substituents in benzene ring with electronegative fluorine atom decreased hypoglycemic activity. Thus, compound 11 in terms of duration and level of a hypoglycemic effect did not exceed, but was only comparable to the glibenclamide in vivo.

The study of 10-(dialkylamino)ethyl-2,3,4,10-tetrahydro-pyrimido[1,2-a]benzimidazoles (12) confirmed their hypoglycemic action [20]. N-Acetamide derivatives of fused tricyclic benzimidazole systems comprising partially hydrogenated imidazole or pyrimidine ring also exhibit hypoglycemic activity. Out of 20 compounds tested nine compounds exceed the reference drug metformin in terms of plasma glucose lowering. Experiments established that substituted amides of 2,3-dihydro-imidazo and 2,3,4,10-tetrahydro-pyrimido[1,2-a]benzimidazolyl-N-acetic acids (13) exhibit different degrees of hypoglycemic activity [21]. Compounds bearing amine or piperidine as the NR¹R² substituent reduced blood glucose level stably for 6 h. The most active compound 14 significantly exceeded the activity of metformin and was comparable in strength to the hypoglycemic action of glibenclamide.

QSAR-guided lead selection

Substructural analysis of the hypoglycemic activity of cyclic guanidine derivatives

One of the problems to solve during drug discovery project is to identify the basic core among chemically diverse structures that is determinant of a particular activity [22]. Classification machine learning techniques have been used successfully to pursue this issue [23], in particular, employing the Bayesian substructural analysis approach [24].

In the course of our study, substructural analysis allowed to determine basic chemical structures responsible for the manifestation of high a hypoglycemic activity among discussed benzimidazole derivatives [25]. Totally, 109 substances of seven chemical classes were studied: 2-aminobenzimidazoles (2-AmBI, Fig. 2); N⁹-imidazo[1,2-a]benzimidazoles (N⁹-ImBI); N¹-imidazo[1,2-a]benzimidazoles (N¹-ImBI); N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazoles (N⁹-DHImBI); N¹-imidazo[1,2-a]benzimidazoles (N¹-DHImBI); N¹,N⁹-imidazo[1,2-a]benzimidazoles (N¹,N⁹-DHImBI) and 2,3,4,10-tetrahydro-pyrimido[1,2-a]benzimidazoles (PrmBI). Ind_R value served as an indicator of hypoglycemic activity and was determined as an average lowering of blood glucose level of experimental animals 4 h after administration of the substance at a 50 mg/kg dose as compared to the control group.

Fig. 2:

QSAR-analysed cyclic guanidine classes.

In terms of Ind_R cluster analysis was performed with k-means method [26]. Twenty-two of the tested compounds belong to a class of a highly active compounds having Ind_R<0.82. As a structural feature, responsible for the manifestation of a high hypoglycemic activity, we have considered the core heterocycles. Analysis of the data (Table 1) allowed us to conclude that N⁹-DHImBI class is the most promising in terms of potency.

Table 1:

Substructural analysis of the influence of the heterocycle core on a high hypoglycemic activity.

Core structure	N a	P h b	P nh c	Pr d	K Pr e
2-AmBI	11	0.1125	0.8875	0.0008	–5
N 9-ImBI	22	0.3056	0.6944	0.0095	–4
N 1- ImBI	3	0.6068	0.3932	0.2824	0
N 9-DHImBI	35	0.5985	0.4015	0.0380	+3
N 1-DHImBI	8	0.5558	0.4442	0.3414	0
PrmBI	30	0.5848	0.4152	0.0851	+2

1. ^aNumber of compounds.

2. ^bFrequency of occurrence of compounds with high hypoglycemic activity.

3. ^cFrequency of occurrence of compounds with non-high hypoglycemic activity.

4. ^dSignificance of differences between P_h and P_nh according to binomial test.

5. ^eInformativity coefficient for differences between P_h and P_nh: 0 – Pr>0.2; 1 – 0.1<Pr≤0.2; 2 – 0.05<Pr≤0.1; 3 – 0.01<Pr≤0.05; 4 – 0.001<Pr≤0.01; 5 – 0.001<Pr; positive values correspond to the positive influence and vice versa.

PrmBI and N¹-DHImBI derivatives can also be considered as a source of hypoglycemic agents, but the influence of this basic structure is not statistically significant enough. However, the introduction of substituents that define activity could result in a high hypoglycemic effect. According to the informativity coefficient 2-AmBI and N⁹-ImBI classes were found to be definitely not promising. No final conclusions could be done for N¹-ImBI class since only three compounds were sampled, which is statistically insufficient.

Therefore, we concluded that N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole class is the most likely to offer highly active hypoglycemic agents.

Substructural analysis of the hypoglycemic activity of N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives

We performed substructural analysis for the 56 experimentally studied N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives [25] aiming to establish the effect of substituents on a high level of hypoglycemic activity (Ind_R<0.82). We considered substituents R¹, R², R³ and type of acid HX as determinants of high activity (Fig. 3). Totally, 52 structural features were analyzed, and their impact on activity level was assessed with informativity coefficient K_Pr. As a result, 19 structural features were defined that contribute to a high hypoglycemic potency (Table 3).

Fig. 3:

N ⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives.

Table 3:

Substructural analysis of N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives (R¹=H).^a

Structural feature	Ph	Pnh	Pr	KPr
	0.2407	0.7593	0.0932	–2
–Et	0.2910	0.7090	0.1828	–1
–Bn	0.7258	0.2742	0.1504	+1
–Me	0.6381	0.3619	0.1903	+1
	0.6381	0.3619	0.1903	+1
	0.7064	0.2936	0.0355	+3
	0.7258	0.2742	0.1504	+1
	0.7258	0.2742	0.1504	+1
	0.7258	0.2742	0.1504	+1
	0.7258	0.2742	0.1504	+1
	0.7258	0.2742	0.1504	+1
	0.2910	0.7090	0.1828	–1
	0.7258	0.2742	0.1504	+1
	0.7258	0.2742	0.1504	+1
	0.2910	0.7090	0.1828	–1
2 HNO3	0.7258	0.2742	0.1504	+1
HCl	0.4171	0.5829	0.1110	–1
2 HBr	0.2910	0.7090	0.1828	–1
I−	0.7142	0.2858	0.0899	+2

1. ^aDesignations are the same as in Table 1.

Sixteen of them have a very slight, statistically insignificant impact on the high activity level. Two features have a little, statistically significant impact: phenoxymethyl in R² position affects adversely, and iodide as an acid residue affects positively. Diethylaminoethyl residue at R³ was considered as a statistically significant determinant of a high level of hypoglycemic activity for N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazoles.

Frequency analysis of physicochemical parameters of N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives

In order to identify additional QSAR that are specific for N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazoles with a high hypoglycemic activity (Ind_R<0.82) we carried out the frequency analysis of a lipophilic, steric and electronic parameters of 56 tested compounds [30].

Lipophilicity was assessed as logP calculated using various fragment approximation methods for entire molecule and certain R substituents separately [31]. Molar refractivity MR served as a steric parameter, which values were calculated using classical additive scheme [32]. Atomic charges Q were calculated with AM1 approach after molecular geometry optimization with MM2 molecular mechanics [33]. A total of 40 physicochemical parameters were calculated.

The impact of each parameter was estimated by substructural analysis algorithm [25] and rated with the informativity coefficient K_Pr. Table 4 collects the most relevant physicochemical features of a high hypoglycemic activity (K_Pr>+2 threshold). Thus, 10 physicochemical parameters were found to reliably determine the presence or absence of a high hypoglycemic activity of the discussed compounds. The data obtained could direct further search for a novel highly active substances in N⁹-DHImBI chemical class.

Table 4:

Frequency analysis of the physicochemical parameters of N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole derivatives.

Parametera	P h	P nh	Pr	K Pr
0.029≤Min(Q+)CS <0.037	0.7834	0.2166	0.0056	+4
60.69≤MR0 <75.04	0.7373	0.2627	0.0146	+3
13.16≤MR3 <27.51	0.7030	0.2970	0.0170	+3
M(Q) 3 ≥0.066	0.7984	0.2016	0.0103	+3
Min(Q + ) CS ≥0.037	0.8267	0.1733	0.0191	+3
Q(N 4 ) CS<–0.232	0.6964	0.3036	0.0326	+3
0.111≤M(Q)2 <0.156	0.6119	0.3881	0.0900	+2
0.030≤M(Q)3 <0.066	0.6863	0.3137	0.0634	+2
0.069≤Min(Q+)3 <0.095	0.6179	0.3821	0.0996	+2
Max(Q + ) CS ≥0.240	0.7394	0.2606	0.0570	+2
Q(Ac R )<–0.213	0.6863	0.3137	0.0634	+2
0.018≤Min(Q+)3 <0.043	0.1478	0.8522	0.0071	–4
−0.006≤M(Q)3 <0.030	0.3986	0.6014	0.0492	–3
Max(Q + ) 3 <0.156	0.2882	0.7118	0.0372	–3
0.014≤Min(Q+)CS <0.021	0.2537	0.7463	0.0124	–3
−0.384≤Q(Het1)3 <−0.299	0.3598	0.6402	0.0890	–2
−0.185≤Q(N4)CS <−0.139	0.2118	0.7882	0.0538	–2

1. ^a M(Q)_i – mean charge; M|Q|_i – mean absolute charge; Max(Q⁺)_i – maximal positive charge; Max(Q⁻)_i – maximal negative charge; Min(Q⁺)_i – minimal positive charge; Min(Q⁻)_i – minimal negative charge; Q(Het) – charge of heteroatom; Q(N₁)_CS – charge of N₁; Q(N₉)_CS – charge of N₉; Q(N₄)_CS – charge of N₄; Q(Benz)_CS – overall charge of benzene ring; Q(Imid₁)_CS – overall charge of inner imidazole ring; Q(Imid₂)_CS – overall charge of outer imidazole ring; Q(Ac_R) – overall charge of acid residue with protonation of R; Q(Ac_Cond) – overall charge of acid residue with protonation of heterocycle.

To sum up, the substructural, median and supremal analyzes of 109 experimentally tested cyclic guanidines rendered N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazole as a promising scaffold for a novel hypoglycemic agents due to the core heterocycle influence and potential to activity increase upon its decoration. Substructure analysis of 56 N⁹-2,3-dihydro-imidazo[1,2-a]benzimidazoles revealed that diethylaminoethyl at N⁹ favors hypoglycemic potency. According to a frequency analysis of the physicochemical parameters the charge on the inner imidazole ring Q(Imid₁)_CS≥−0.109 is a significant sign of a high hypoglycemic activity.

Thus, the results of computational analysis taken together led us to a conclusion that compound RU-254 (diabenol, 9-diethylaminoethyl-2,3-dihydro-imidazo[1,2-a]benzimidazole dihydrochloride, Fig. 4) represents all the attributes of a high hypoglycemic activity, and thereby is the most promising cyclic guanidine derivative that should be pursued to further evaluate its antidiabetogenic properties.

Fig. 4:

Diabenol – compound RU-254.

Experimental and clinical evaluation of RU-254 (diabenol)

The effect of diabenol on insulin secretion was subsequently studied in vivo. We have shown that diabenol increases plasma insulin concentration in lean cats more than four times as compared to the baseline at the 10^th min of experiment reflecting recovery of the first phase of insulin secretion. This effect was maintained for 60 min, bringing insulin to an average of 80.7 mU/ml. At 60–90-min period insulin concentration decreased twice, and 30 min later it has dropped another 33.3%. On the 120^th min the hormone concentration in the blood of cats returned to the original values [34].

Dose-dependent hypoglycemic effect of diabenol was determined in a single and repeated administration experiments. In rat studies ED₃₀ of diabenol matches that of gliclazide and it exceeds glibenclamide 1.5 times. Rabbits treated with diabenol 2 h after administration produced more pronounced hypoglycemia than animals treated with gliclazide. After 4 h, diabenol exceled gliclazide by 10.1%. It has been found that duration of the diabenol hypoglycemic action is independent from experimental animal species and lasts 20–22 h, which is similar to the duration of sulfonylureas action.

It is important to note that diabenol has no effect on a blood glucose in dogs with removed pancreas, but combination of diabenol with exogenous insulin has more pronounced and long lasting effect than administration of insulin alone [34], suggesting that diabenol improves peripheral insulin sensitivity. Another evidence is the increase of glycogen content in insulin-dependent tissues along with increased glycolysis and reduced glycogenolysis, which was observed in rat alloxan-induced diabetes model.

Moreover, diabenol exhibits marked antithrombogenic action that was revealed in platelet-dependent thrombosis models [35]. In particular, diabenol efficiently blocked ADP-induced platelet aggregation, which is a key mechanism of thrombosis in T2DM, exceeding the effect of gliclazide. Inhibition of platelet aggregation induced with arachidonic acid and low concentrations of collagen was also observed along with a decrease in malonic dialdehyde formation [36]. This could lead to assumption that diabenol inhibits production of TXA₂ in platelets.

Experiments have shown the hemorheological activity of diabenol. We observed the charge increase of erythrocyte membranes, which causes a decrease in aggregation of cells during blood circulation, and increased microviscosity of erythrocyte membranes, which increases their plasticity. Erythrocyte aggregation induced with La³⁺ was reduced to a similar extent as with reference membranotropic drug pentoxifylline [37].

No toxic and cancerogenic effects were observed during the chronic administration of diabenol in NMRI mice with low cancer incidence, HER-2/neu transgenic mice, and LIO rats. Importantly, diabenol in NMRI mice retards the development of age-related extral functions disorders and increases life span of animals. Drug inhibited the occurrence of spontaneous tumors, reduced incidence of malignant lymphomas, hindered emergence and development of a colon cancer induced by 1,2-dimethylhydrazine in rats. Rats treated with diabenol developed colon adenocarcinoma (induced by carcinogen) with exophytic type of growth, higher degree of differentiation and less invasion depth. That is, tested dosage of diabenol has experimentally proven antitumor and geroprotective effects [38].

Taking into account results of experimental studies we advanced diabenol to the clinical trials. During phase 1–2 it was established that hypoglycemic and hemorheological activities contribute to antidiabetic action of diabenol rendering it more efficient than the reference drug gliclazide [39]. Diabenol effect manifested in conditions of insulin resistance, dysfunction of pancreatic β-cells, increased blood viscosity syndrome, i.e. typical complex of T2DM condition. It was found that diabenol restores physiological profile of insulin secretion, reduces HbA_1c level, inhibits platelet aggregation and normalize lipid metabolism.

Phase 3 clinical trial was designed as randomized controlled comparative study of the efficacy, tolerability, and safety. Glidiab (gliclazide) was chosen as a reference drug. Upon chronical administration diabenol reduced fasting and postprandial blood glucose, reduced glycated hemoglobin level by 1.1%, and increased postprandial plasma insulin level by the end of the 3^rd month [39]. Also, diabenol decreased platelets aggregation, increased erythrocytes deformability and reduced their aggregation, thus normalizing coagulation hemostasis. A significant decrease in blood viscosity was noted at a different shear rates that simulate blood flow characteristics of large and small diameter vessels. Diabenol decreased thrombogenic potential of blood by inhibition of platelet aggregation induced by ADP and arachidonic acid, which confirms earlier experimental data. During the investigation it was found that diabenol also affects coagulation hemostasis, normalizing both extrinsic and intrinsic pathways of blood clotting.

According to the clinical monitoring of patients, it could be concluded that administration of diabenol during 3 months had no adverse or toxic effects evident from biochemical blood and urine analysis but resulted in the sufficient hypoglycemic effect. Diabenol could be recommended for including in clinical practice as a hypoglycemic agent for the treatment of type 2 diabetes complicated with angiopathies.

Conclusions

A class of cyclic guanidine derivatives is a valuable and promising source of a potential antidiabetic agents. Experimental evaluation with subsequent computer-aided lead selection allowed us to identify the novel clinical candidate RU-254 – diabenol, which was found to stimulate insulin secretion resulting in a sustainable hypoglycemia in diabetic animals. It also inhibits platelet aggregation and demonstrates antitumor and geroprotective properties. Pronounced antidiabetic activity and tolerability of diabenol was confirmed in phase 3 clinical trial.