Design, synthesis, and discovery of 5-((1,3-diphenyl-1H-pyrazol-4-yl)methylene)pyrimidine-2,4,6(1H,3H,5H)-triones and related derivatives as novel inhibitors of mPGES-1

Abstract

Human mPGES-1 has emerged as a promising target in exploring a next generation of anti-inflammatory drugs, as selective mPGES-1 inhibitors are expected to discriminatively suppress the production of induced PGE₂ without blocking the normal biosynthesis of other prostanoids including homeostatic PGE₂. Therefore, this therapeutic approach is believed to reduce the adverse effects associated with the application of traditional non-steroidal anti-inflammatory drugs (tNSAIDs) and selective COX-2 inhibitors (coxibs). Identified from structure-based virtue screening, the compound with (Z)-5-benzylidene-2-iminothiazolidin-4-one scaffold was used as lead in rational design of novel inhibitors. Besides, we further designed, synthesized, and evaluated 5-((1,3-diphenyl-1H-pyrazol-4-yl)methylene)pyrimidine-2,4,6(1H,3H,5H)-triones and structurally related derivatives for their in vitro inhibitory activities. According to in vitro activity assays, a number of these compounds were capable of inhibiting human mPGES-1, with the desirable selectivity for mPGES-1 over COX isozymes.

Graphical Abstract

Rational molecular design, followed by synthesis and in vitro activity assays for evaluating both the potency and selectivity, has led to the discovery of a set of novel, potent and selective mPGES-1 inhibitors.

In the eicosanoid pathway, arachidonic acid (AA) is converted to prostaglandin H₂ (PGH₂) by the action of cyclooxygenases (COX-1 and COX-2). PGH₂ serves as a common precursor for various biologically active prostanoids, such as thromboxane A₂ (TXA₂), PGD₂, PGI₂, PGF_2α, and PGE₂, depending on different distal synthases.^{1, 2} Among these prostanoids, PGE₂ is well recognized as an important inflammatory mediator. PGE₂ is isomerized from PGH₂ catalyzed by three distinct synthases, including microsomal prostaglandin E₂ synthase-1 (mPGES-1), mPGES-2, and cytosolic prostaglandin E₂ synthase.³ Unlike the other two constitutively expressed enzymes, the expression of mPGES-1, similar to that of COX-2, is highly inducible in response to pro-inflammatory stimuli.⁴

As two generations of anti-inflammatory drugs, traditional non-steroidal anti-inflammatory drugs (tNSAIDs) and coxibs represent the mainstream for the treatment of inflammation-related symptoms by either non-selectively inhibiting COX isozymes,⁵ or selectively inhibiting COX-2,⁶ respectively. However, both categories shut down the biosynthesis of all downstream prostanoids and, so, their application is associated with considerable adverse effects. The tNSAIDs trigger gastrointestinal (GI) ulceration because of the interference with COX-1-derived protective function in GI tract;⁷ coxibs, as specific COX-2 inhibitors, on the other hand, break the internal balance of vasodilative PGI₂ and vasoconstrictive TXA₂ and thus result in cardiovascular risk.⁸ Since PGE₂ is the major inducible PG in inflammation, inhibiting mPGES-1 is believed as a promising therapeutic approach in the development of the next generation of anti-inflammatory drugs.⁹

In the previous study, we reported the discovery of (Z)-5-benzylidene-2-iminothiazolidin-4-one derivative 1 (Figure 1A) as a human mPGES-1 inhibitor with moderate potency (IC₅₀ = 3.5 μM) through structure-based virtual screening.^{10, 11} Based on the simulated binding mode of 1 with crystal structure (PDB ID: 4BPM,¹² Figure 1B) of human mPGES-1, we redesigned the core scaffold and synthesized benzylidenebarbituric acid derivatives (2, with IC₅₀ = 622 nM, as an example, Figure 1A) as mPGES-1 inhibitors. The barbituric acid was introduced in order to maintain and possibly enhance the polar interaction with the enzyme at the active site. Flexible alkoxy side chain was attached to the central benzene ring so as to fit the size of the hydrophobic pocket surrounded by I32, G35, L39, Y130, T131, L135, and A138. As predicted by molecular docking, the barbituric ring interacts with S127 by forming a hydrogen bond (HB) between the barbituric carbonyl group and hydroxyl group of serine, while the central benzene ring with alkoxy substitution occupies the hydrophobic groove where the long hydrocarbon “tail” of PGH₂ locates (Figure 1C).

Figure 1.

Molecular structures of the lead 1, benzylidenebarbiturc acid derivative 2 and the designed scaffold 3 and their binding with human mPGES-1. (A) Ligand structures; (B) binding with the lead 1; (C) binding with 2; (D) binding with 3.

While developing the benzylidenebarbituric acid derivatives, we carefully analyzed the binding mode of these compounds with mPGES-1, and noticed that there is still substantial unoccupied area in the active site, especially the small hydrophobic pocket around the central benzene ring of 2 and above the cofactor glutathione (GSH) of the enzyme. We then decided to introduce pyrazole into the core scaffold not only because of its existence in many bioactive molecules, but also its versatility for multi-functionalization. We kept the barbituric acid “head” because of the importance of the hydrogen bond between carbonyl and hydroxyl group of S127. Substituents on pyrazole-1 and 3 positions were expected to occupy the hydrophobic pockets wrapping the hydrocarbon “tail” of PGH₂ and above GSH. Thus, a series of 5-((1,3-diphenyl-1H-pyrazol-4-yl)methylene)pyrimidine-2,4,6 (1H,3H,5H)-trione derivatives and other structurally related compounds were designed and synthesized. The docking study of the simplest scaffold 3 (Figure 1A) with mPGES-1 revealed that 3 binds in a similar region in the active site as 2 in the binding complex (Figure 1D). The strategies of improving the inhibitory efficacy in light of the binding complex include the modification on barbituric acid moiety for maintaining and strengthening a hydrogen bond with S127, and the substitution on 1- and/or 3-phenyl in order to enhance the hydrophobic interaction with the nonpolar groove. As expected, a number of these compounds were active against human mPGES-1 and selective for mPGES-1 over COX isozymes.

The synthesis of this series of compounds followed a straightforward multi-step protocol, as shown in Scheme 1. 4-Alkoxyacetophenone (4b~4g), obtained from the reaction of 4-hydroxyacetophenone and alkyl bromide,¹³ or acetophenone (4a) was condensed with 4-chlorophenylhydrazine in reflux ethanol containing 5 % glacial acetic acid.^{14, 15} The ethylidene hydrazine (5a~5g) was formed as precipitate at room temperature and filtered off. The next step was Vilsmeier-Haack-Arnold ring closing formylation,¹³ by treating 5a~5g with POCl₃/DMF. The produced 1H-pyrazole-4-carbaldehyde intermediate (6a~6g) was coupled with barbituric acid or 2-thiobarbituric acid in refluxing EtOH/H₂O (4:1, v/v) to afford the final product (7a~7g or 8a~8g).¹⁶

Scheme 1.

Reagents and conditions: (a) K₂CO₃ (2.00 equiv.), DMF, 80 °C; (b) 5 % glacial AcOH in EtOH, reflux; (c) POCl₃ (4.00 equiv.), DMF, 0 °C~60 °C; (d) EtOH/H₂O (4:1, v/v), reflux.

The length and shape of the aliphatic side chain were investigated in the SAR study. We fixed the substituent at pyrazole-1-position as 4-chlorophenyl and variated the side chain on 3-phenyl. From the in vitro data shown in Table 1, it was observed as compared to that without a side chain (7a and 8a), compounds with linear side chains (7b~7f and 8b~8f) were generally more potent against human mPGES-1, whereas benzyl substitution (7g and 8g), however, did not improve the inhibitory potency. Linear side chains with 4 or 6 carbons yielded compounds with highest potency, whereas longer side chains, such as octyl or decyl, did not show a more potent inhibition. Notably, compounds with 2-thiobarbituric acid “heads” were generally more potent as compared to those with barbituric acid ones. We also changed the substituent in pyrazole-1-position from 4-chlorophenyl to phenyl group. In this case, 4c was used as starting substituted acetophenone. Followed the similar protocol as outlined in Scheme 2, 11 and 12 were prepared. These compounds (11 and 12) were slightly less potent than those with 4-chlorophenyl substituent (7c and 8c, respectively).

Table 1.

Structural-activity relationship (SAR) on the substitution of central pyrazole ring

Comp.	R2	X	Y	IC50 (nM)a for mPGES-1
7a	H	Cl	O	337±85
7b	C2H5O	Cl	O	265±96
7c	nC4H9O	Cl	O	169±41
7d	nC6H13O	Cl	O	285±65
7e	nC8H17O	Cl	O	361±51
7f	nC10H21O	Cl	O	294±83
7g	BnO	Cl	O	598±142
8a	H	Cl	S	561±192
8b	C2H5O	Cl	S	95±16
8c	nC4H9O	Cl	S	56±10
8d	nC6H13O	Cl	S	52±15
8e	nC8H17O	Cl	S	92±19
8f	nC10H21O	Cl	S	93±14
8g	BnO	Cl	S	797±160
11	nC4H9O	H	O	212±34
12	nC4H9O	H	S	92±20

^aData are expressed as means ± SD (standard deviation) of single determinations obtained in triplicate.

Scheme 2.

Reagents and conditions: (a) 5 % glacial AcOH in EtOH, reflux; (b) POCl₃ (4.00 equiv.), DMF, 0 °C~60 °C; (c) EtOH/H₂O (4:1, v/v), reflux.

With these (2-thio)barbituric acid derivatives in hand, as depicted in Scheme 3, we broadened the structural diversity with pyrazole core by coupling 1H-pyrazole-4-carbaldehyes (6b and 6d) with various activated methylene compounds such as malononitrile, 2-cyanoacetic acid and 2,4-thiazolidinedione. As shown in Table 2, compounds with malononitirle and 2-cyanoacetamide “heads” (13b, 13c and 14b, 14c) were not very active against human mPGES-1, whereas those with 2-cyanoacetic acid (13a and 14a) showed submicromolar potency. It was noted that compounds 13f and 14f, obtained from the coupling of 6b and 6d with 2,4-thiazolidinedione N-acetic acid, were capable of inhibiting human mPGES-1 with low naonomolar potency (IC₅₀ = 41 ± 5 nM and 36 ± 11 nM, respectively).

Scheme 3.

Reagents and conditions: (a) NH₄OAc (2.00 equiv.), glacial AcOH, 108 °C.

Table 2.

The SAR on the polar head

Comp.	R		IC50 (nM)a for mPGES-1
13a	C2H5		283±83
13b	C2H5		n.d.b (51%±10%)c
13c	C2H5		n.d. (32%±21%)
13d	C2H5		1,590±560
13e	C2H5		1,040±290
13f	C2H5		41±5
14a	nC6H13		83±34
14b	nC6H13		n.d. (64%±1.4%)
14c	nC6H13		n.d. (14%±14%)
14d	nC6H13		1,390±300
14e	nC6H13		1,730±670
14f	nC6H13		36±11

^aData are expressed as means ± SD (standard deviation) of single determinations obtained in triplicate.

^bn.d. = not determined.

^cThe value in the parenthesis refers to the % inhibition of the compound at a concentration of 10 μM against the mPGES-1 (IC₅₀ values were determined only for the compounds that showed ≥70% inhibition).

To further understand the SAR of these synthesized compounds, we selected compounds 7c, 7e and 14f and used the AutoDock Vina program¹⁷ as the molecular docking tool to investigate their binding modes with human mPGES-1. 14f was chosen for further docking study because it is the most active one with the scaffold (possessing a carboxylic acid polar “head”) for the mPGES-1 inhibitors in Table 2. 7c and 7e were chosen from compounds in Table 1 with another scaffold (barbituric acid derivatives). Of the barbituric acid derivatives in Table 1, we are mainly interested in the compounds with Y = O, because the compounds with Y = S are less soluble in water. Within all of the barbituric acid derivatives with Y = O in Table 1, 7c is the most potent one for human mPGES-1. 7e was also chosen for comparison with 7c because they are different only in the length of the side chain (R²).

The predicted binding modes are shown in Figure 2. The substituted phenyl group on pyrazole-1-position occupies the hydrophobic groove where the long hydrocarbon side chain of PGH₂ locates. The other substituted phenyl group on pyrazole-3-position fits into the small hydrophobic pocket above GSH. Since this pocket is small, a bulky group on pyrazole-3-position is not favorable as reflected in the inhibitory data that 7e (with longer octyl side chain) is an inferior inhibitor as compared with 7c (with a shorter butyl side chain). The substituents on pyrazole-4-position attaches on the surface of the protein. When the barbituric acid is present, the carbonyl group on the barbituric ring forms a hydrogen bond with the hydroxyl group of S127. . While replacing the barbituric acid with 2,4-thiazolidinedione N-acetic acid, however, the carboxyl group forms hydrogen bonds with the NH groups of R52 and H53. So, 14f is a potent inhibitor against human mPGES-1.

Figure 2.

Molecular structures of 7c, 7e and 14f and their binding with human mPGES-1. (A) Compound structures; (B) binding with the lead 7c; (C) binding with 7e; (D) binding with 14f.

For the in vitro evaluation of these compounds, we first conducted the single-concentration screening at 10 μM against human mPGES-1. Compounds that showed significant inhibition (≥70%) were tested further for their IC₅₀ values against human mPGES-1. The protocol for the protein preparation and in vitro activity assays were the same as described previously.^{10, 11, 18, 19} Further, the inhibitory activity against COX isozymes was also evaluated for some of the more promising compounds (with IC₅₀ < 100 nM against human mPGES-1). As shown in Table 3, at a concentration as high as 100 μM, compounds 8b~8f, 12, 13f, 14a and 14f inhibited COX-1/2 for less than 20 %. So, these compounds are highly selective for the mPGES-1 over COX-1/2.

Table 3.

Inhibition against COX-1/2

Compound	% Inhibition of COX-1/2 at 100 μMa
8b	1.0±4.5
8c	4.4±8.6
8d	0.7±5.4
8e	0±0.2
8f	11±16
12	16±4.1
13f	6.8±6.4
14a	19±0.1
14f	1.2±29

^aThe % inhibition of the compound at a concentration of 100 μM against the COX-1/2 (mixed COX-1 and COX-2). The enzyme mixture contained equal amounts of COX-1 and COX-2 in terms of their enzyme activities. In this way, when a compound can significantly inhibit either COX-1 or COX-2, it will show the significant inhibitory effects against the mixed COX-1 and COX-2.

In summary, in light of the binding complex of both the lead and the benzylidenebarbituric acid scaffold with human mPGES-1, we designed and synthesized a novel series of 5-((1,3-diphenyl-1H-pyrazol-4-yl)methylene)-pyrimidine-2,4,6 (1H,3H,5H)-triones and other structurally related derivatives.²⁰ These compounds were evaluated in vitro for their inhibitory potency against human mPGES-1 and selectivity over COX-1/2, leading to discovery of various potent and selective inhibitors of human mPGES-1. The most potent one is 14f (IC₅₀ = ~36 nM against human mPGES-1) without significant inhibition against COX-1/2.