BIM-46174 fragments as potential ligands of G proteins

Fragments of BIM-46174 were synthesized and investigated as Gα_q inhibitors.

Abstract

The 5,6,7,8-tetrahydroimidazo[1,2-a]pyrazine derivative BIM-46174 has received attention as Gα_q inhibitor. We conducted structural reductions to monocyclic and bicyclic substructures to explore the chemical space of BIM fragments and to gain insights into the pharmacophore of BIM-type Gα_q inhibitors. Two piperazin-2-one-containing fragments and a small library of bicyclic lactams featuring fused pyrazine and diazepine rings were synthesized and evaluated. The results of a second messenger-based cellular assay indicate that the entire BIM structure is required for efficient Gα_q inhibition.

Seven transmembrane G protein-coupled receptors (GPCRs) represent the largest family of cell-surface molecules and play a key role in signal transmission.1,2 Agonist binding to the receptor stabilizes its active conformation, which stimulates cytoplasmic heterotrimeric guanine nucleotide-binding proteins (G proteins), composed of three distinct subunits, α, β, and γ. The capability of acting as molecular switches, thus transducing extracellular signals via GPCRs into intracellular signal cascades, makes G proteins vitally important.3–5 Whereas GPCRs are major drug targets, the intracellular G proteins have received less attention for the design and development of new tool compounds or drug candidates. In case of complex pathologies, such as cancer, which involve multiple receptors and their associated pathways, the manipulation of signalling at the post-receptor level appears to be a particular promising therapeutic approach. Based on the different α subunits, G proteins are subdivided into four families, Gα_s, Gα_i/o, Gα_q and Gα_12/13. Among Gα_q proteins, the ubiquitously expressed isoforms Gα_q and Gα₁₁ are most prominent.6–8 Gα_q proteins customarily activate phospholipase C-β, leading to hydrolysis of membrane-bound phosphatidylinositol-4,5-bisphosphate to diacylglycerol, which in turn activates protein kinase C, and to myo-inositol 1,4,5-trisphosphate (IP3) which initiates the release of calcium ions from the sarcoplasmic reticulum.4 Only few compounds are available that can modulate Gα_q protein activity, including the two natural products YM-254890 and FR900359 (UBO-QIC) and the synthetically accessible compound BIM-46174 (1, Fig. 1) and its disulfide, BIM-46187. Such modulators of G protein activity represent valuable pharmacological tools to investigate G-protein-mediated signalling.8 The cyclic depsipeptides YM-254890 and FR900359 can be isolated from the fermentation broth of Chromobacterium sp. QS3666 and from the leaves of the plant Ardisia crenata, respectively. Both natural products are able to selectively block heterotrimeric G proteins of the Gα_q family.6–12 The two BIM molecules have also been employed experimentally for G protein inhibition.12–19 They have been previously reported as pan-G protein inhibitors able to inhibit the four families of Gα proteins to a similar extent.13,15 Detailed investigation on the mode of action then revealed that BIM molecules preferentially silence Gα_q signalling in a cellular context-dependent manner by trapping Gα_q in the empty pocket conformation, permitting GDP exit and preventing GTP entry.20 Even though these BIM compounds are not as selective and potent as YM-254890 or FR900359, their potential for chemical modifications should be higher and can be realized by employing combinatorial chemistry, e.g. to overcome some undesired properties of BIM compounds, such as low solubility and toxicity.

Fig. 1. Structure of BIM-46174.

This study was aimed at exploring the chemical space of BIM-46174 fragments as potential ligands of G proteins. We synthesized and evaluated monocyclic and bicyclic parts of the BIM monomer in a fragment-based approach,21–24 adopted to clarify which structural unit of the BIM molecule is responsible for its activity.

In order to elucidate essential substructures, a first attempt was directed towards the Western part of BIM. Accordingly, we maintained the piperazine ring and its cysteinyl substituent and devised a synthetic route to fragment 2 (Scheme 1). Starting from Boc-protected (S)-cyclohexylalanine, we used ethanolamine as the source of one nitrogen and two carbons needed for the piperazine ring formation.25,26 To enable a Mitsunobu reaction, N-benzylethanolamine was introduced through a uronium salt-mediated amide coupling. The resulting intermediate 11 was Boc-deprotected and subjected to the cyclization with diisopropyl azodicarboxylate and triphenylphosphine using the Mitsunobu protocol. A Birch reduction of 13 with sodium in liquid ammonia led to the intermediate 14 with preserved stereochemistry at C-3.27,28 Seibel et al. reported on a similar Birch debenzylation of 2-oxopiperazines which occurred without racemization.25,26 Moreover, we obtained single diastereomers, i.e.16 and 2, after attachment of the (R)-configured cysteine. Compound 16, obtained through a carbodiimide-promoted amide formation, was subjected to the concomitant release of the amino and thiol group under acidic conditions. Triisopropylsilane, acting as a hydride donor, was used as a scavenger for the S-deprotection to trap the trityl cation. The final compound 2 was isolated conveniently through precipitation of the hydrochloride salt in the presence of the additive dithiothreitol which prevented oxidation of the product. For the design of 3, the structure was further reduced and the cyclohexylmethyl group in 2 was removed. This compound 2 was obtained analogously from 2-oxopiperazine (15) in two steps.

Scheme 1. Synthesis of 4-(cysteinyl)piperazin-2-one fragments 2 and 3a. Reagents and conditions: (a) 2-(benzylamino)ethanol, HATU, Et₃N, abs. CH₂Cl₂, rt, 18 h, N₂, 80% (11); (b) TFA/CH₂Cl₂ (1 : 4), rt, 2 h, 39% (12); (c) DIAD, PPh₃, abs. THF, rt, 5 h, 35% (13); (d) Na, NH₃, abs. THF, –78 °C, 20 min, 54% (14); (e) Boc-l-Cys(Trt)-OH, DCC, DIPEA, abs. CH₂Cl₂, rt, 48 h, Ar, 74% (16), 93% (17); (f) (i-Pr)₃SiH, TFA, abs. CH₂Cl₂, rt, 18 h, Ar; (g) DTT, abs. EtOAc, abs. MeOH, 1 M HCl in EtOAc, rt, 3 h, Ar, 18% (2), 46% (3).

The second attempt to BIM fragments was focused on the Eastern, bicyclic core (Scheme 2). A Davidson-type synthesis was employed by first esterifying Cbz-cyclohexylalanine (18) with 2-bromoacetophenone followed by treatment of the O-acyl α-hydroxy ketone 20 with ammonium acetate in toluene. The mechanism of the imidazole formation includes two nucleophilic attacks of ammonia, one at the ketone carbonyl and the second, after an O-to-N 1,4-acyl shift and tautomerisation, at the aldehyde carbon, followed by cyclocondensation.29 In crystal structures of similar imidazoles, the nitrogen in position 1 was protonated, as depicted.29–32 Next, compound 22 was reacted with ethyl 2-bromoacetate. It was intended to confirm the regioselectivity of this N-alkylation since it is relevant not only for structure type 24, but also for the final scaffold of this route (4) and beyond. The concomitant deprotection and cyclization afforded product 4. Its structure was corroborated by X-ray diffraction analysis; see ESI† (Fig. S2). Hence, for the first time, it was verified by crystal structure analysis that the alkylation of 2-alkyl-4-aryl-imidazoles takes place at the NH nitrogen in position 1. The glycine-derived product 5 was prepared accordingly.

Scheme 2. Synthesis of imidazo[1,2-a]pyrazin-6(5H)-one fragments 4 and 5. Reagents and conditions: (a) 2-bromoacetophenone, K₂CO₃, abs. DMF, rt, 4 h, 82% (20), 61% (21); (b) NH₄OAc, abs. toluene, reflux, 3 h, 92% (22), 33% (23); (c) ethyl 2-bromoacetate, Cs₂CO₃, abs. DMF, rt, 3.5 h, 83% (24), 72% (25); (d) Pd/C, H₂ (1 atm), abs. MeOH, 24 h, 81% (4), 42% (5).

The synthesis of 4 and 5 was carried out similarly to a published route towards imidazo[1,2-a]pyrazin-6(5H)-ones which have been further converted to efficient antimalarials. The most potent agents bear a substituted 3-anilino group, introduced via the corresponding 3-bromo derivative, an 8-dimethyl substitution, a glycyl or methylalanyl group at 7-position and methylene replacing the carbonyl unit.33–36

The bicyclic 5,6,7,8-tetrahydroimidazo[1,2-a]pyrazine system is also accessible through other synthetic routes, e.g. a three-component coupling of 2-aminopyrimidine isonitriles and aldehydes followed by catalytic reduction.37 Starting from 2-aminoketones and 2-ethoxy-2-iminoacetates, imidazole-2-carboxamides were obtained and the six-membered ring was closed by double alkylation with 1,2-dibromoethane to yield RIP1 kinase inhibitors.38

In order to produce ring-expanded analogues, we used the intermediates 22 and 23 (Scheme 3) and performed the N-alkylation with ethyl 3-bromopropionate. After cleavage of the Cbz group of 26 and 27, mixtures of the non-cyclized amino esters and the desired products were observed, clearly because of the hindered cyclization to a 7-membered ring. However, prolonged heating in the presence of catalytic p-toluenesulfonic acid promoted the ring closure to 8,9-dihydro-5H-imidazo[1,2-a][1,4]diazepin-7(6H)-ones 6 and 7. Other representatives of this heterocyclic scaffold have been obtained from imidazoles with 1-propargyl and 2-hydroxmethyl substituents which were converted with sulfonyl azides via ketenimines, cyclic sulfonimidates and, through rearrangement, N-sulfonyllactams.39

Scheme 3. Synthesis of imidazo[1,2-a][1,4]diazepin-7(6H)-one fragments 6 and 7. Reagents and conditions: (a) ethyl 3-bromopropionate, Cs₂CO₃, abs. DMF, rt, 3.5 h, 88% (26), 47% (27); (b) Pd/C, H₂ (1 atm), abs. MeOH, 24 h; (c) abs. EtOAc, TsOH, reflux, 16 h, 55% (6), 31% (7).

Another enlargement of the 6-membered ring of compound 4 comprised the introduction of a methylene unit between the lactam nitrogen and the chiral carbon. The required amino acid (Scheme 4), i.e. (R)-3-amino-2-(cyclohexylmethyl)propanoic acid (35), was synthesized following a Seebach protocol.40 The acid chloride 29, derived from 3-cyclohexylpropionic acid (28), was employed for chiral derivatization with an Evans' oxazolidinone auxiliary. The resulting 30 forms a nucleophilic titanium enolate in the presence of titanium tetrachloride and triethylamine,41 and was reacted with N-(chloromethyl)benzamide (32), which was accessible from the hydroxyl precursor 31. The auxiliary induces a diastereoselective alkylation from the si face of the cis enolate, giving rise to an enantiomerically pure product.42,43 Hydrolytic removal of the auxiliary under mild basic conditions with lithium hydroperoxide, followed by debenzoylation with hydrochloric acid and N-protection provided 36.

Scheme 4. Synthesis of (R)-3-amino-2-(cyclohexylmethyl)-propanoic acid 36. Reagents and conditions: (a) SOCl₂, abs. CH₂Cl₂, reflux, 18 h, 98% (29); (b) (S)-4-isopropyloxazolidin-2-one, n-BuLi, abs. THF, –78 °C to rt, 2 h, 85% (30); (c) SOCl₂, abs. CH₂Cl₂, rt, 77% (32); (d) TiCl₄, Et₃N, abs. CH₂Cl₂, 0 °C, 2 h, 79% (33); (e) H₂O₂ (30%), LiOH × H₂O, THF/H₂O (4 : 1), 0 °C, 1 h, 52% (34); (f) concd HCl/AcOH/H₂O (2 : 1 : 1), reflux, 48 h, 81% (35); (g) Cbz-OSu, Et₃N, THF/H₂O (3.5 : 1), rt, 18 h, 88% (36).

To prepare a further Eastern fragment (Scheme 5), the carboxylic group of the N-protected β²-amino acid 36 was esterified and converted to an imidazole through Davidson-type reaction. N-1-Alkylation with ethyl bromoacetate yielded the intermediate 42. Its lactamisation after N-deprotection occurred smoothly. This 7-membered ring closure to 8, in contrast to the formation of 6 (Scheme 3), is facilitated by the altered electron densities, increased at the nitrogen and decreased at the carbonyl carbon. When Cbz-β-alanine (37) was applied, the corresponding achiral route gave the 8,9-dihydro-5H-imidazo[1,2-d][1,4]diazepin-6(7H)-one 9.

Scheme 5. Synthesis of imidazo[1,2-d][1,4]diazepin-6(7H)-one fragments 8 and 9. Reagents and conditions: (a) 2-bromoacetophenone, K₂CO₃, abs. DMF, rt, 4 h, 75% (38), 65% (39); (b) NH₄OAc, abs. toluene, reflux, 3 h, 43% (40), 39% (41); (c) ethyl 2-bromoacetate, Cs₂CO₃, abs. DMF, rt, 3.5 h, 65% (42), 79% (43); (d) Pd/C, H₂ (1 atm), abs. MeOH, 24 h, 77% (8), 99% (9).

As reported, BIM-46174 reduced calcium release in human melanoma cells, induced by the GPCR agonist endothelin-1. It inhibited cAMP production in MCF-7 cancer cells after treatment with the Gα_s activator choleratoxin, but not in cells pretreated with the adenylate cyclase activator forskolin.13 The compound was shown to silence Gα_q signalling of the muscarinic M1 receptor over Gα_s signalling of E-type prostanoid receptors 2 and 4 (EP2/4) or Gα_i signalling of serotonin 5-HT receptors in a CHO cell background. In human embryonic kidney (HEK) 293 cells, a Gα_q inhibitory effect of BIM-46174 was observed upon muscarinic M3 receptor activation by the agonist carbachol (CCh), in contrast to Gα_s and Gα_i signalling determined at EP2/4 and DP2 (formerly CRTH2) receptors, respectively, which was not affected.20 To evaluate the fragments of BIM-46174 in experiments with HEK293 cells, an assay based on M3 receptor activation was chosen. Hence, we measured Gα_q inhibition in a competitive immunoassay for the quantitative determination of myo-inositol 1-phosphate (IP1). IP1, a downstream metabolite of IP3, accumulates in cells upon Gα_q activation through the M3 agonist CCh.20 IP1 was measured after 2 hours incubation of test compounds followed by 35 min CCh stimulation.

Apart from BIM-46174 (1), none of the tested compounds (2–9) showed the potential to inhibit CCh-stimulated IP1 accumulation in HEK293 cells at a concentration of 100 μM (Fig. 2). Selected fragments with sufficient solubility were investigated at higher concentrations (Fig. S1, ESI†). A weak, but concentration-dependent inhibition of the Gα_q-mediated IP1 accumulation was observed for 5 and 7. An intra-assay control was performed with cells transfected to express a mutated phospholipase C β3 (PLCβ3) variant producing IP1 in a Gα_q-independent manner. Here, fragments 5 and 7 did not affect the PLCβ3 activity indicating a specific, Gα_q-mediated inhibition of the IP1 accumulation.

Fig. 2. Gα_q-dependent IP1 formation in carbachol-stimulated HEK293 cells pretreated with 100 μM of BIM-46174 (1) and BIM fragments (2–9). Data are means ± s.e.m. of three independent experiments. ***P < 0.001 compared to w/o using Dunnett's multiple comparisons after one-way ANOVA.

Furthermore, the toxicity of all compounds was measured in a CellTiter-Blue viability assay (Fig. 3). Compared with BIM-46174 (1), the fragments 2–9 revealed to be much less toxic in HEK293 cells. Thus, although Gα_q inhibition per se is not linked to cellular toxicity and both properties are possibly independent,6 efficient Gα_q inhibition and cellular toxicity were only observed for the unfragmented BIM molecule.

Fig. 3. Viability of HEK293 cells treated with 100 μM BIM-46174 (1) and BIM fragments (2–9). Data are presented as mean ± s.e.m. of three independent experiments. **P < 0.01 compared to w/o using Dunnett's multiple comparisons after one-way ANOVA.

Our results of the IP1 assay strongly indicate that the entire BIM structure is required for efficient Gα_q inhibition. Noteworthy, on the one hand, when the hetero-biphenyl system was removed, the activity got lost (1versus2). On the other hand, compound 4 comprising the heterobicyclic system and the original phenyl substituent of BIM, did not show inhibitory activity either, suggesting that the cysteinyl moiety is essential. In agreement with this assumption, fragments 6 and 8 with the seven-membered 1,4-diazepinone ring lack activity, as compound 4 did. Our data did not provide information to estimate the influence of the cyclohexylmethyl group, present in fragments 2, 4, 6, 8 and absent in fragments 3, 5, 7, 9.

As a conclusion, we identified both the aromatic Eastern substructure and the polar, redox-reactive Western part of BIM to be requisite for notable Gα_q inhibition. It can be expected that after cellular uptake, through thiol–disulfide exchange reactions, e.g. with glutathione, involving the thioredoxin–glutaredoxin system, BIM may undergo an intracellular activation and interact with protein thiol groups.44–46 A possible covalent interaction with cysteine residues of G proteins (e.g. Cys330 or Cys144, conserved in all Gα_q proteins) was discussed with respect to an impaired GTP binding in the switch regions and a compromised motion of G protein domains.20 The fragment-based approach reported herein, should now be implemented in a future combinatorial chemistry effort towards a library of BIM analogues.

Conflicts of interest

There are no conflicts to declare.