Design and Synthesis of Bis-amide and Hydrazide-containing Derivatives of Malonic Acid as Potential HIV-1 Integrase Inhibitors

HIV-1 integrase (IN) is an attractive and validated target for the development of novel therapeutics against AIDS. In the search for new IN inhibitors, we designed and synthesized three series of bis-amide and hydrazide-containing derivatives of malonic acid. We performed a docking study to investigate the potential interactions of the title compounds with essential amino acids on the IN active site.


Introduction
Therapeutic protocols for the treatment of HIV infection are mainly based on the combined use of reverse transcriptase, protease, and more recently, of cell fusion and entry inhibitors. Although drugs targeting reverse transcriptase and protease are in wide use and have shown effectiveness, the rapid emergence of resistant variants, often cross-resistant to the members of a given class, limits the efficacy of existing antiretroviral drugs. Therefore, it is critical to develop new agents directed against alternate sites in the viral life cycle.
In this context, HIV-1 integrase (IN), the enzyme that mediates an obligatory step in the viral replication process by catalyzing the integration of viral cDNA into the host genome, represents a validated target for the development of new drugs against HIV-1 infection [1]. Because IN does not have a human homologue, it represents one of the most promising targets in AIDS research.
In the past several years, numerous compounds with diverse structural features have been reported as IN inhibitors [2,3]. In particular a number of compounds bearing a β-diketo acid moiety (DKA, I, Figure 1) were discovered as new selective and potent inhibitors [4], and some of them have emerged as the most promising lead in anti-HIV-1 IN drug discovery.
Recently, the IN inhibitor MK-0518 (Merck & Co., Inc., II, Figure 1), a DKA-like compound, has been approved in therapy [5]. In the light of these promising results it is important to design new and potent IN inhibitors. To identify novel and unified pharmacophore required for activity we selected and formally combined the main structural motifs of I and II together to the hydrazide functionality of compounds III and IV (Figure 1), previously reported as a new class of selective IN inhibitors having antiviral activity [6]. With this in mind, we designed two sets of symmetrical (1) and asymmetrical (2) bisamide derivatives of malonic acid (Figures 1 and 2). Although several synthetic and biological studies on DKAs and DKA-based compounds have been reported [1,4], the mechanism responsible for inhibition of the IN still remains uncertain. It is believed that the β-diketo acid pharmacophoric motif could be involved in a functional sequestration of one or both divalent metal ions in the enzyme catalytic site, to form a ligand-M 2+ -IN complex, which blocks the formation of the IN-DNA complex by competing with the target DNA substrate [1]. In light of such hypothesis we focused on structures bearing a tautomeric moiety, as in I-DKA (Figure 3), with the aim that the cation coordination at the IN catalytic site is favoured. The possibility to obtain a pharmacophoric fragment suitable to generate tautomers and complexes with metal ions was supported by the previoulsy reported behaviour of β-ketoamide and hydrazide systems [7][8][9][10][11].
On this basis, a new set of molecules with structures 3a-f ( Figure 4) was designed and synthesized considering a) bulkiness of substituents, and b) the substitution of an amidic motif with a hydrazidic one, hopefully suitable for coordination with the enzyme catalytic site and/or with metallic cofactors.

Scheme 2. Preparation of bis-amides 2b-h.
Reagents and conditions: i) DMTMM, THF, r.t., 24 h. Desymmetrization of dimethyl ester 5 to monoamide 7 was accomplished via aminolysis with one equivalent of aniline 4a to give 8, which was separated by filtration and readily hydrolized with two equivalents of Na 2 CO 3 in water (Scheme 3, method a) [15]. Alternatively, treatment of the commercially available monobenzyl ester 9 with aniline 4a in the presence of DMTMM afforded 10 that was converted to the desired 7 by hydrogenolysis at r.t. (Scheme 3, method b). The use of DMTMM has been reported to be particularly useful for the formation of amides starting from carboxylic acids and aromatic amines [16].

Preparation of hydrazides.
Similar to the afore-mentioned synthetic strategy for bis-amides 1a-h, hydrazides 3a-e were synthesized by condensation of the amido ester 11 with the corresponding hydrazine 12a-e in ethanol [17] or chlorobenzene [18] under reflux, although a slight modification in the procedure for 3f was required (Scheme 4).  [19] by nucleophilic substitution of aniline 4a on ethyl malonyl chloride (EMC). As depicted in Scheme 5, condensation of 11 with 12c afforded the expected 3c along with the dianilide 1a (23%), which was characterized by comparison with data reported in the literature [19,20]. Moreover, treatment of 11 with 12e gave the bis-hydrazide 14 as the major product (14 = 60% vs 3e = 21%). The formation of the bis-anilide 1a (along the formation of diethylmalonate 6) can be rationalized via an intermolecular condensation of 11 in its tautomeric form to give the adduct 15, which can cyclize intramolecularly to afford the four membered cyclic transition state 16 followed by structural collapse as end game ( Figure 5) [19]. The synthesis of symmetrical bis-anilides 1a-like has previously been reported. It consists of treatment of diethyl esters of the malonic acids and anilines in solvent free conditions (170-180 °C), that it has been used for the preparation of 1a-h (as depicted in Scheme 1), or under reflux in high boiling point solvents such as DMF [19]. In the example reported herein, the formation of 1a is facilitated because 11 (enolic tautomer) can allow the ester carbonyl to be in conjunction with the aniline system causing further reduction of its electronic density and therefore favouring the formation of the adduct 15.

Reagents and conditions:
On the other hand, the new compound 14 is likely to be formed from transamidation of the desired 3e as a discrete excess of hydrazide 12e (0.5 eq) is used. Full characterization of 14 was accomplished by NMR, mass spectrometry and elemental analysis. Compound 3f was obtained in 89% yield by condensation of hydrazide 3a with commercially available α-tetralone 13 in ethanol under reflux.
Finally, analysis of the 1 H-NMR and 13 C-NMR spectra of all described compounds (recorded in DMSO-d 6 or CDCl3 + DMSO-d 6 mixture) revealed the presence of the methylene between the two carbonyls, and no extra-OH signals are detected. Moreover, D 2 O exchange involved only amidic NH (3a-d and 11) and hydrazidic NH (3a-d). This indicates that such systems seem to exist, in organic solvents, only as diketo tautomers.
In this study, AutoDock 3.0.5 was used because it utilizes a fully flexible ligand in its docking algorithm (although it is still docked to a rigid protein) and because it has been shown to successfully reproduce many crystal structure complexes [28]. We selected compounds 1a, 2h, 3a, 3c, 3d, 3e, and the reference compounds I-Phe-DKA and III, and built them in their neutral form (only I-Phe-DKA was calculated as carboxylate). Moreover, three putative tautomers (3a 1-3 , Figure 7) for the hydrazide 3a were postulated and included in this preliminary study. The results of clustered docking runs with the most favourable free binding energy are given in Table 1   According to docking results, the amino acid residues involved in the binding of title compounds located near the catalytic site were as follows: Asp64, Cys65, Thr66, His67, Glu92, Asp116, Ser119, Asn120, Asn155, Ly156. Several of these were considered to be very important for the activity of IN and some have been previously shown to play a role in substrate binding [29,30,23]. In general, different ligands showed different binding modes with some overlapping features, which were predicted as potential H-bonds and van der Waals interactions.
The estimated free binding energy values (ΔG bind ) of the docked positions, expressed in kcal/mol, indicated favourable interactions and tight binding with key aminoacid residues on the active site of IN. However, with ΔG bind = -7.78 kcal/mol, I-Phe-DKA displayed better energy results than the other compounds, as expected on the basis of its reported potency against IN [31]. High free binding energy values have also been obtained for the reference compound III. We also found that III binds to Cys65 and chelates Mg 2+ , thus revealing the already observed accommodation [6] at the IN active site ( Figure  6).
Moreover, compounds 2h, 3c and 3d, showed similar free binding energy, when compared with the reference compound III (ΔG bind = -6.86, -6.55, -6.56 and -6.78 kcal/mol for 2h, 3c, 3d and III, respectively). Interestingly, the three keto-enol tautomers of 3a shared different free binding energy results (ΔG bind = -4.41, -5.56, and -5.61 kcal/mol for 3a 1 , 3a 2 , and 3a 3 , respectively). In fact, while 3a 1 showed a significantly lower binding energy, for 3a 2 , and 3a 3 a similar behaviour in terms of both free binding energy and binding modes was observed. This could be due to the fact that the keto-enolic tautomers could potentially be able to establish a coordination bond with Mg 2+ ion through an hydroxy-keto system (for 3a 2 ) or to strongly bind (for 3a 3 ) with important residues on the active site of IN as displayed in Figure 7.

Conclusions
We have synthesized novel symmetric and asymmetric bis-amides and hydrazides derivatives of malonic acid as prototypes of IN inhibitors and built a unified pharmacophore to help us in rational design of future inhibitors. Docking studies were performed on title and reference compounds to predict their binding interaction with the active site of IN. Studies are in progress to optimize their potency and physicochemical properties.

General
Anhydrous solvents and all reagents were purchased from Sigma-Aldrich, Merck, Acros or Carlo Erba. Reactions involving air-or moisture-sensitive compounds were performed under a nitrogen atmosphere using oven-dried glassware and syringes to transfer solutions. Melting points (m.p.) were determined using an Electrothermal melting point or a Köfler apparatus and are uncorrected. Infrared (IR) spectra were recorded as thin films or nujol mulls on NaCl plates with a Perkin-Elmer 781 IR or 983 spectrophotometers and are expressed in ν (cm -1 ). Nuclear magnetic resonance spectra ( 1 H-NMR and 13 C-NMR) were determined in CDCl 3 /DMSO-d 6 (in 3/1 ratio) or DMSO-d 6 and were recorded on a Varian XL-200 (200 MHz) or a Varian VXR-300 (300 MHz). Chemical shifts (δ scale) are reported in parts per million (ppm) downfield from tetramethylsilane (TMS) used as internal standard. Splitting patterns are designated as follows: s, singlet; d, doublet; t, triplet; q, quadruplet; m, multiplet; brs, broad singlet; dd, double doublet. The assignment of exchangeable protons (OH and NH) was confirmed by addition of D 2 O. Analytical thin-layer chromatography (TLC) was carried out on Merck silica gel F-254 plates. For flash chromatography Merck Silica gel 60 was used as stationary phase with a particle size 0.040-0.063 mm (230-400 mesh ASTM). Elemental analyses were performed on a Perkin-Elmer 2400 spectrometer, and were within ±0.4% of the theoretical values.

General procedure for the synthesis of symmetric bis-amides (1a-h)
Dimethyl malonate (for 1a-e and h) or diethyl malonate (for 1f-g) (5.6 mmols) and the corresponding aromatic amine (11.0 mmols) were mixed under an argon atmosphere in a 25 mL twoneck round-bottom flask fitted with a reflux condenser and a magnetic bar. The mixture was stirred at 185 °C for 3.5-5 h. The solids obtained were triturated from EtOH and isolated by filtration. The products were characterized as reported below. General procedure for the synthesis of asymmetric bis-amides 2b-h. In a two-neck 50 mL roundbottom flask were introduced, under an argon atmosphere, an aromatic amine (2.6 mmols) and the monophenylamide of malonic acid 7 (0.46 g, 2.6 mmols) in anhydrous THF (20 mL). The reaction mixture was stirred at r.t. for 15 minutes. Then, 4-(4,6-dimethoxy-[1,3,5]-triazyn-2-yl)-4-methylmorpholine chloride (DMTMM, 0.70 g, 2.6 mmols) were added and the solution was stirred at r.t. for 24 h. The solvent was evaporated under vacuo and the solid residue obtained was washed with water, an aqueous saturated solution of NaHCO 3 , a 1.0 M aqueous solution of HCl and brine. Finally, the solid was washed with diethyl ether and filtered.

Preparation of N-phenylmalonamic acid (7).
Method a) In a two-neck 100 mL round-bottom flask fitted with a condenser and stir bar were introduced under an argon atmosphere aniline 4a (6.5 g, 7.0 mmols) and dimethylmalonate (13.8 g, 10.0 mmols). The reaction mixture was vigorously stirred at 190 °C for 30 minutes and then cooled to r.t. and diluted with ethanol (40 mL). The solid formed was filtered through Büchner and the filtrate was diluted with water (50 mL) in order to facilitate the precipitation of diamide still in solution that was separated by filtration. The filtrate was extracted with dichloromethane (DCM) and the combined organic layers were washed with 1N HCl, dried over Na 2 SO 4 , filtered and the solvent was evaporated in vacuo to give a yellow oil (5.08 g, 2.6 mmols, 37 % yield); TLC (ethyl acetate/petroleum ether = 7/3). The obtained compound was transferred in a 250 mL round bottom flask fitted with a condenser and a stirring bar and was suspended in a solution of Na 2 CO 3 (5.29 g, 5.0 mmols) in water (100 mL). The mixture was refluxed for 30 min, cooled down with a slush bath and acidified by slow addition of 37 % HCl until a white precipitate was formed.

Method b).
To a two-neck 500 mL round bottom flask fitted with a condenser and a stirring bar containing anhydrous THF (150 mL) were added under a nitrogen atmosphere monobenzyl ester 9 (6.0 g, 30.0 mmols) and aniline 4a (2.73 mL, 30.0 mmol). The solution was stirred at r.t. for 15 min. and then DMTMM (8.3 g, 30.0 mmols) dissolved in anhydrous THF (50 mL) were added. The reaction mixture was stirred at r.t. for 24 h and then the solvent was evaporated in vacuo to give a white residue that was dissolved in DCM, washed with a saturated aqueous solution of NaHCO 3 , water, 1N HCl and brine. The organic layer was dried over Na 2 SO 4 , filtered and the solvent was evaporated in vacuo to afford 10 as a white solid that was triturated from petroleum ether and DCM (5.2 g, 52.4 mmols). Yield = 62 %); m.p. = 62 -63 °C.
The amido ester 10 was transferred under a nitrogen atmosphere into a 100 mL two-neck flask fitted with a stirring bar containing anhydrous MeOH (50 mL). Then 5 % Pd/C (0.52 g, 0.2 mmols of Pd/0.01 eq) was added and the mixture was hydrogenated at r.t. and atmospheric pressure for 4 h. A second portion of 5 % Pd/C (0.52 g, 0.2 mmols of Pd/0.01 eq) was added and the reaction stirred for 12 h. The mixture was treated with nitrogen, filtered, and the filtrated was evaporated in vacuo to give 7 as a white solid (2.96 g, 16.5 mmols). Yield = 86 %; m.p. = 132 °C (lit. [15] 132 °C).  (14). 4-Trifluoromethyl-2-hydrazinopyrimidine 12e (0.45 g, 2.53 mmols) was added to a solution of 11 (0.35 g, 1.69 mmols) in chlorobenzene (15 mL) and the mixture was stirred at reflux for 25 h. The solution was cooled down to r.t. and the precipitate formed was filtered and washed with chlorobenzene. Then, the solid was washed with ethyl acetate to give 14 as a white solid that remained on the filter, and evaporation of the organic layers in vacuo afforded 3a as a yellow solid.

Molecular Modeling
Model compounds I-Phe-DKA, III, 2a, 2h, 3a 1 , 3a 2 , 3a 3 , 3c, 3d, and 3e were constructed with standard bond lengths and angles from the fragment database with MacroModel 6.0 [38] using a Silicon Graphics O2 workstation running on IRIX 6.3. Sybyl 6. 2 (2001) [39] was used as graphic platform. The atomic charges were assigned using the Gasteiger-Marsili method [40]. Minimization of structures was performed with the MacroModel/BachMin 6.0 program using the AMBER force field. Extensive conformational search was carried out using the Monte Carlo/Energy minimization [41] for all the compounds considered in the study (E i -E min < 5 Kcal/mol, energy difference between the generated conformation and the current minimum). Final minimization of the structures was performed with Sybyl 6.2 by using Tripos force field. Docking calculations were performed on Linux workstation.
Subunit A of IN core domain in complex with 1-(5-chloroindol-3-yl)-3-hydroxy-3-(2H-tetrazol-5yl-propenone) (5CITEP; PDB 1QS4) was used for all docking studies. The missing residues at positions 141-144 in this subunit were incorporated from monomer B of the IN structure PDB 1BIS after superimposition of the backbones of residues 135-140 and 145-150, as previously reported [21][22][23]. The atomic charges for the protein were assigned using the Kollman United method [42]. Docking was performed with AutoDock version 3.05 [28] using the empirical free energy function and the Lamarckian protocol [43]. Mass-centered grid maps were generated with 80 grid points for every direction and with 0.375 Angstroms spacing by the AutoGrid program for the whole protein target.
Random starting position on the entire protein surface, random orientations and torsions were used for the ligands. The distance-dependent dielectric permittivity of Mehler and Solmajer was used for the calculation of the electrostatic grid-maps. 100 independent docking runs were carried out for each ligands. The cluster analyses were computed with a cluster tolerance by less than 1.5 Å in positional root-mean-square deviation with AutoDock Tools 1.4.6 [44].