Structure-activity-relationship study of N-acyl-N-phenylpiperazines as potential inhibitors of the Excitatory Amino Acid Transporters (EAATs): improving the potency of a micromolar screening Hit is not truism

The excitatory amino acid transporters (EAATs) are transmembrane proteins responsible for the uptake of (S)-glutamate from the synaptic cleft. To date, five subtypes EAAT1-5 have been identified for which selective inhibitors have been discovered for EAAT1 and EAAT2. By screening of a commercially available compound library consisting of 4,000 compounds, N-acyl-N-phenylpiperazine analog (±)-exo-1 was identified to be a non-selective inhibitor at EAAT1-3 displaying IC50 values in the mid-micromolar range (10 μM, 40 μM and 30 μM at EAAT1, 2 and 3, respectively). Subsequently, we designed and synthesized a series of analogs to explore the structure-activity-relationship of this scaffold in the search for analogs characterized by increased inhibitory potency and/or EAAT subtype selectivity. Despite extensive efforts, all analogs of (±)-exo-1 proved to be either inactive or to have least 3-fold lower inhibitory potency than the lead, and furthermore none of the active analogs displayed selectivity for a particular subtype amongst the EAAT1-3. On the basis of our findings, we speculate that (±)-exo-1 binds to a recess (deepening) on the EAAT proteins than a well-defined pocket.


Background
In the central nervous system (CNS), the excitatory amino acid transporters (EAATs) are transmembrane proteins responsible for the uptake of (S)-glutamate (Glu) from the synaptic cleft. Five subtypes have been identified, named EAAT1-EAAT5 in humans and GLAST, GLT-1, EAAC1, EAAT4 and EAAT5, respectively, in rodents. ) While EAAT5 is found exclusively in the retina, subtypes EAAT1-4 are expressed differentially within the CNS with respect to brain regions as well as at the cellular level: EAAT1 and EAAT2 are expressed primarily on astrocytes, but EAAT2 is also found in neurons, astrocytes and oligodendrocytes. (Lauriat et al. 2007) Subtype EAAT3 is distributed predominantly in postsynaptic neuronal sites, (Nieoullon et al. 2006) whereas EAAT4 is distributed in Purkinje cells as well as in the cerebral cortex. (Massie et al. 2001) Discovery of subtype selective ligands for the EAATs has attracted much attention over the past decade, ) the latest being the disclosure of UCPH-101 as first subtype selective EAAT1-inhibitor ( Figure 1). Erichsen et al. 2010;Huynh et al. 2012,ab).
Although phenylpiperazines are promiscuous hits in high-throughput screenings (HTS) and a frequent core skeleton in marketed drugs, (Millan et al. 2001;Fragasso et al. 2006;Weisberg et al. 2007) we were motivated to explore the structure-activity-relationship (SAR) of this new class of EAAT inhibitors. A conventional medicinal chemistry analysis of (±)-exo-1 suggests that the amide functionality, the aniline nitrogen, the phenyl ring and the trifluoromethyl group may play key roles in binding of this class of EAAT inhibitors. Consequently, the chemical structure can be broken down into three fragments: a core skeleton being the acyl-phenylpiperazine scaffold and two substituents being the trifluoromethyl-and the bicyclo [2.2.1]heptanyl group ( Figure 1). The SAR study was designed as to study the influence of one of the two substituents ( Figure 1) individually including the stereochemical organization around the α-carbonyl carbon.
The SAR study commenced by investigation of the influence of bicycle [2.2.1]heptanyl group on the EAAT inhibitory activity. The stereochemical configuration of the α-cabonyl carbon was addressed by the synthesis of endoconformer (±)-endo-1 (Table 1) from commercially available (±)-endo-carboxylic acid and N-(3-trifluoromethylphenyl)piperazine 4, the latter prepared by a palladium-catalyzed amination of commercially available piperazine (Scheme 1). (Nishiyama et al. 1998) Furthermore, the racemic and diastereomeric mixture (±)-endo-exo-1 was prepared from the corresponding acid (±)-endo-exo-6 obtained from oxidation of commercially available (±)-endo-exo-bicyclo [2.2.1]heptanylmethyl alcohol ((±)-endo-exo-5) (Scheme 2) using KMnO 4 and K 2 CO 3 in H 2 O. Isolation of (±)-endoexo-6 turned out to be difficult for which reason it was used directly in the next step. (Gudipati et al. 1993) To search for the optimal bulkiness of the lipophilic substituent, larger as well as smaller rigid hydrophobic ring-systems were introduced (2.1−2.5). Moreover, analogs 2.6−2.10 comprising alkyl group of varying length and bulkiness were designed to explore the effect of increased flexibility of this substituent on ligand binding. Furthermore, analogs 2.11-2.17 address if the bicyclo [2.2.1]heptane could be substituted for an aromatic moiety, whereas the analogs 2.18-2.21 were designed to explore the distinct substitution for a hydrophobic group. The synthesis of piperazine analogs 1 and 2.1−2.21 was carried out by amidation of 4 using the respective carboxylic acids, acid chlorides, benzenesulfonyl chloride and benzyl carbonochloridate afforded the corresponding amides in moderate to good yields (Scheme 1). The rationally designed 3-trifluoromethylphenylpiperazine analogs were supplemented by commercially available analogs 2.22-2.31, as a quick way of expanding the SAR into the chemical space beyond rational guidance. Finally, the importance of the amide functionality was explored by the synthesis of carbamate 2.32 by acylation of 4 with carbonochloridate, sulfonamide 2.33 by treatment of 4 with phenylsulfonyl chloride, amine 2.34 by reduction of (±)-endo-exo-1 with Figure 1 Chemical structures of the EAAT1-selective inhibitor UCPH-101, Erichsen et al. 2010) and EAAT2-selective inhibitors DHK, (Jensen & Bräuner-Osborne 2004) DPAG, (Sagot et al. 2008) and WAY-213613 (Dunlop et al. 2005).
We then turned to the design of analogs for investigation of the influence on EAAT inhibitory activity of the chemical nature of the trifluoromethyl group as well as its position on the phenyl ring. A series of 12 analogs were included in the SAR study, all wherein the (±)-endo-exo-bicyclic [2.2.1]-acyl group was conserved (analogs 3.1−3.12, Table 2): Simplification of the chemical structure by depletion of the trifluoromethyl group provides analog (±)-endo-exo-3.1, while shifting the 3-trifluoromethyl group to the 4-or 2-positions affords analogs (±)-endoexo-3.2 and (±)-endo-exo-3.5, respectively (Table 2 and Scheme 2). The latter two analogs were supplemented by commercially available analogs (±)-endo-exo-3.3, (±)-endoexo-3.4 and (±)-endo-exo-3.6. Continuing the design stage, substitution of the 3-trifluoromethyl group for a chloride, hydroxyl-, cyano-and methoxy group, provides analogs (±)-endo-exo-3.7−3.10 respectively (Table 2), whereas 2,4-difluorophenyl analog (±)-endo-exo-3.11 was included due to readily available starting materials. Analog (±)-endoexo-3.12 could be obtained from commercial suppliers and thus included with the notion that it comprises an N-diphenylmethyl group, which is indeed chemically distinct from the N-3-trifluoromethylphenyl group and furthermore the connecting nitrogen will be protonated at physiological pH=7.4. The analogs were synthesized starting from the appropriate phenylpiperazine and (±)-endo-exo-6 under standard coupling conditions (TBTU, DIPEA in DMF) for the amide formation to afford the target compounds in moderate yields (Scheme 3) (Balalaie et al. 2007).

Pharmacological characterization
In total, 54 piperazine analogs 2.1−2.40 and 3.1−3.12 were characterized pharmacologically at stable HEK293 cells expressing human EAAT1−3 in a [ 3 H]-D-aspartate uptake assay, (Jensen & Bräuner-Osborne 2004) and the results are summarized in Table 1 and Table 2. The endo-isomer 1 (endo-isomer) displayed inhibitory activities at EAAT1−3 comparable with those of the exo-isomer 1 (lead structure) (IC 50 = 14 μM, 32 μM and 10 μM vs. 10 μM, 40 μM and 30 μM, respectively). In line with this, endo-exo 1, which is a 1:1 ratio of endo/exo moiety displayed IC 50 values at EAAT1−3 of IC 50 = 14 μM, 42 μM and 14 μM, respectively. Usually, such findings would lead to the conclusion that the bicyclo-[2.2.1]-heptanyl group occupies a promiscuous hydrophobic pocket, which could be optimized for increased potency. However, upon increasing or decreasing the hydrophobic bulk and/or flexibility, a clear drop in potency was observed (analogs 2.1−2.17, Table 1). Except for analogs 2.3 and 2.11, which displayed only a 5−15 fold drop in inhibitory potency across the subtypes, all of these analogs would be characterized as inactive (IC 50 >100 μM or >300 μM, Table 1). These findings could open up for the hypothesis that the pocket is indeed not hydrophobic but instead hydrophilic in nature. Upon binding of the hydrophobic alkane group, water molecules are forced out and ligand binding is entropically driven rather than enthalpically. Thus analogs 2.18−2.31, which comprise a hydrophobic group, could be potential inhibitors. However, none of these displayed any inhibitory activity at EAAT1−3 (IC 50 >100, >300 or >1000 μM, Table 1). Continuing the characterization, neither the carbamate 2.32 nor the sulfonamide 2.33 analog displayed inhibitory activity at EAAT1−3, and likewise all amines 2.34−2.40 were found to be inactive at all three subtypes.
The pharmacological results for the twelve analogs 3.1−3.12, which address the influence of substituent 2 ( Figure 2) on inhibitory activity at EAAT1−3 are summarized in Table 2. While it was not surprising that removal or repositioning of the 3-trifluoromethyl group (analog 3.1, 3.2 and 3.5, respectively) resulted in loss of inhibitory activity, the further nine analogs 3.3, 3.4, 3.6−3.12 were also inactive or displayed at least a 3-fold lower inhibitory activity than (±)-exo-1.
at EAAT1−3 but only the endo diastereomer (±)-endo-1 displayed inhibitory potency in the mid-micromolar range comparable to that of the lead structure (±)-exo-1. The remaining analogs were inactive or at least three fold weaker inhibitors at EAAT1−3 than the lead, none of them displaying signs of subtype-selectivity. Given the structural diversity of the analogs characterized pharmacologically, we speculate if the lead structure (±)-exo-1 adheres to a recess (deepening) in the surface of the protein rather than binds in an organized way to a well-defined pocket.

Experimental section
All commercially available reagents were used without further purification. THF was distilled over sodium/benzophenone, Et 2 O was dried over neatly cut sodium and dichloromethane was dried over 3 Å molecular sieves. All solvents were tested for water content using a Karl Fisher apparatus. All reactions involving dry solvents or sensitive agents were performed under a nitrogen atmosphere, and glassware was dried prior to use. All reactions were monitored by analytical thin-layer chromatography (TLC, Merck silica gel 60 F 254 aluminum sheets General procedure a: synthesis of amides using O-benzotriazole-1-yl-N,N,N',N'-tetramethyluronium tetrafluoroborate (TBTU) as coupling reagent To a suspension of the appropriate phenylpiperazine analog (0.33 mmol), the carboxylic acid (0.40 mmol) and TBTU (0.43 mmol) in dry DMF (4 mL) under an N 2 atmosphere, was added DIPEA (1.32 mmol) and reaction mixture was stirred for 20 h at rt. The reaction mixture was quenched with brine (5 mL) and extracted with dichloromethane (3 × 20 mL). The combined organic phases were washed with H 2 O (20 mL) and brine (20 mL) and dried over anhydrous Na 2 SO 4 . After concentration in vacuo, the crude product was purified by column chromatography on silica gel in accordance with details described for the analog.
General procedure B: synthesis of amides using acid chlorides To a suspension of 1-(3-(trifluoromethyl)phenyl)piperazine (4) (0.33 mmol) in dry dichloromethane (5 mL) at 0°C under a N 2 atmosphere was added Et 3 N (0.91 mmol). The reaction mixture was stirred for 10 min at 0°C, then the appropriate acid chloride (0.48 mmol) was added and stirring continued for 30 minutes at rt. The reaction mixture was quenched with sat. NH 4 Cl (5 mL) and extracted with dichloromethane (3 × 20 mL). The combined organic phases were washed with H 2 O (20 mL) and brine (20 mL) and dried over Na 2 SO 4 . After concentration in vacuo, the crude product was purified by column chromatography on silica gel in accordance with details described for the analog.