Synthesis of Azide-Tagged Library of 2,3-Dihydro-4-Quinolones

Abstract

We describe the assembly of a 960-member library of tricyclic 2,3-dihydro-4-quinolones using a combination of solution-phase high-throughput organic synthesis and parallel chromatographic purification. The library was produced with high efficiency and complete chemo and diastereoselectivity by diversification of an azide-bearing quinolone via a sequence of [4+2] cycloadditions, N-acylations and reductive aminations. The azide-functionalization of this library is designed to facilitate subsequent preparation of fluorescent or affinity probes, as well as small-molecule/surface conjugation.

Assembly of nitrogen-containing heterocyclic libraries has been a major focus of high-throughput organic synthesis due to the prevalence of such chemotypes among bioactive compounds.¹ Representative examples include 1,4-benzodiazepines,² 1,4-dihydropyridines,³ dihydropyrimidines,⁴ pyrrolidines⁵ and purines.⁶ However, despite a number of innovative synthetic approaches to such small-molecule libraries reported in the past, the existing compound collections occupy only a small fraction of biogenic chemical space.⁷ Thus, synthesis of new heterocyclic libraries to enable discovery of new bioactive chemotypes continues to be of significant importance. We describe the assembly of a 960-member library of tricyclic dihydroquinolones, which was produced with high efficiency and excellent chemical purity by a combination of miniaturized solution-phase synthesis and high-throughput chromatographic purification. Azide-functionalization of this library was designed to facilitate subsequent cellular target identification and construction of small-molecule microarrays.

Bicyclic 2,3-dihydro-4-quinolone subunit (1, Fig. 1) is found in a number of bioactive small molecules, which include the acronycine family of alkaloids with potent anticancer properties (2),⁸ antagonists of CRTH2 receptor (3),⁹ as well as a class of antimitotic agents (4).¹⁰ Such diverse level of functionalization of 2,3-dihydro-4-quinolone moiety promoted us to examine this structural motif as an attractive platform for the assembly of the corresponding heterocyclic library, which was aimed at preserving favorable physicochemical properties of the resulting compounds in order to enable new lead discovery.

Figure 1.

The structure of 3,4-dihydro-4-quinolone (1), which is found in several bioactive chemotypes, including antitumor alkaloids (2), CRTH2 antagonists (3) and antimitotic agents (4).

We selected quinolone A (Scheme 1) as the starting point for introduction of molecular diversity. Indeed, the nitrogen atom can be readily functionalized by N-acylation. The vinylogous amide moiety can enable efficient fusion of additional carbocyclic rings using the [4+2] annulation developed by Beifuss and co-workers.¹¹ Furthermore, the aromatic ring can be readily equipped with an azidomethyl group, which would enable subsequent functionalization of each library member via alkyne-azide [3+2] cycloaddition.¹² The assembly process was designed to begin with a series of [4+2] cycloadditions of quinolone A with enones B to give the initial set of tricyclic amines C, which would be subsequently diversified via N-acylation using readily available acid chlorides D to give the resulting diketones E. Chemoselective and diastereoselective reductive amination of diketones E with amines F would be employed as the final diversity-generating step to deliver the target heterocyclic library G. High-throughput synthesis employing all possible combinations of 10 enones B, 8 acid chlorides D and 12 amines F was expected to deliver 960 tricyclic dihydroquinolones.

Scheme 1.

Synthetic Strategy to Azide-Functionalized Library of Tricyclic 3,4-Dihydro-4-quinolones

Our initial objectives were two-fold. First, we intended to validate the generality and efficiency of the synthetic route shown in Scheme 1. Second, we aimed to develop a simple protocol that would enable production of all final compounds in high chemical purity. To this end, azido quinolone 5 (Fig. 2A), which was readily prepared from 4-nitrobenzyl alcohol, was first subjected to the [4+2] cycloaddition with trans-3-nonen-2-one using the protocol developed by Beifuss and co-workers (TESOTf, 2,6-lutidine, CH₂Cl₂, 20 °C).¹¹ Protodesilylation of the resulting TES enol ether with TFA, followed by subsequent Alloc deprotection (Pd(PPh₃)₄, morpholine) afforded tricyclic diketone 6, which was next N-acylated with benzoyl chloride to give amide 7. The final reductive amination step required differentiation of the two ketones in 7, as well as high level of diastereoselection. Initially, we examined the use NaBH(OAc)₃, which promoted reductive amination with high chemoselectivity but with only moderate diastereoselectivity (dr: 8:2-9:1), which was not acceptable for a final step of the library production. Examination of a number of other reducing agents revealed that this problem could be circumvented by using a bulkier reagent NaBH(OEh)₃ (Eh = 2-ethylhexanoate),¹³ which produced the required amine as a single diastereomer detected by 500 MHz ¹H NMR. Using this protocol, we evaluated the reductive amination of diketone 7 with 12 commercially available primary amines (Fig. 2B). To facilitate parallel chromatographic purification, all reactions were performed on 2-4 mmol scale to deliver the corresponding products (8-19) in good chemical yields (61-71%), high diastereomeric ratios (>95:5) and excellent chemical purities (Fig. 2D).

Figure 2.

Validation Studies. (A) General reaction sequence. (B) Structures of amines used for reductive amination. (C) Purity analysis of 12 purified products by thin-layer chromatography (TLC). (D) ¹H NMR spectra, structures and yields for 5 selected compounds. Yields were determined by 500 MHz ¹H NMR spectroscopy using an internal standard.

The library production began with a selection of enones B to be employed for the first diversification step of the process (Fig. 3). While a range of unsaturated carbonyl compounds successfully participated in the Beifuss [4+2] cycloaddition, only 3-substituted enones proved to be viable for the production of the final library as other substitution patterns were not tolerated at a later stage of the synthesis. We selected 10 readily available enones B (Fig. 3B), which were employed to produce a set of corresponding diketones C following in situ protodesilylation of the TES enol ether and Alloc deprotection. For the second diversification step, we selected a set of seven acid chlorides D to be used for N-acylation of amines C (Fig. 3C), as well as the parent Alloc-protected amine. Each reaction proceeded efficiently in the presence of Et₃N as a base to deliver the required amides E, setting the stage for the completion of the library synthesis.

Figure 3.

Assembly of Azide-Tagged Small-Molecule Library. (A) Reaction sequence. (B) Structures of 10 enones employed for the first diversification step. (C) Structures of 8 acid chlorides employed for the second diversification step. (D) Structures of 12 amines employed for the final diversification step. (E) Representative, fully analytically characterized library members.

The final reductive amination was carried out using NaBH(OEh)₃ as a reducing agent to enable high diastereoselectivity of this transformation. Each of the 80 diketones E was treated with 12 amines F to produce tricyclic dihydroquinolones G on 4.5 μmol scale, followed by parallel preparative TLC purification. Initial analysis revealed that all 960 reactions proceeded successfully to deliver final library members in high chemical purity (see Supporting Information). The purity and efficiency was further quantified by 500 MHz ¹H NMR characterizations of 120 randomly selected library members. In addition, five diketones, five amides, and five amines were subjected to full analytical characterization (Fig. 3F). While the efficiency of the reductive amination ranged from 42% to 75% after silica gel purification, this protocol provided sufficient amounts of material (1.0-1.5 mg of final products) for preparation of DMSO stock solutions to enable 200-500 subsequent high-throughput biological screens. While the selection of this reaction scale greatly facilitated high-throughput purification, standard LCMS purification methods could be readily employed for production of larger amount of material if desired.

Azide-alkyne [3+2] cycloaddition has emerged as a powerful, general and bioorthogonal strategy for covalent molecular conjugation.¹² The azide functionalization of dihydroquinolone library was designed to explore attachment of affinity probes, fluorophores and radiolabels to final library members. Alternatively, a similar approach could be used for surface immobilization of this small-molecule library.¹⁴ Such functionalization would enable development of new high-throughput screens and facilitate cellular target identification of active compounds discovered in phenotypic assays. Treatment of a representative azide 25 with alkyne 26 in the presence of Cu(MeCN)₄PF₆ and TBTA afforded the expected [3+2] cycloadduct 27 in 62% yield (Scheme 2).¹⁵ This experiment validated that the cycloaddition protocol is efficient, fully compatible with a range of functional groups present in dihydroquinolones and could be potentially employed for subsequent derivatization of the entire chemical library.

Scheme 2.

Cu-Catalyzed Azide-Alkyne Cycloaddition

In closing, we have assembled a 960-member library of tricyclic 2,3-dihydro-4-quinolones using a combination of miniaturized solution-phase high-throughput organic synthesis and parallel chromatographic purification. Importantly, this approach enabled rapid validation of synthetic sequences and building blocks for library synthesis, which often represents the bottleneck in the process of generating new chemical libraries. Second, parallel chromatographic purification of each individual library member enabled rapid production of all final compounds in sufficient quantity to enable hundreds of cell-based or target-based high-throughput screens. Broad biological evaluation of this library, as well as the construction of small-molecule microarrays, is currently in progress. Results of these studies will be reported in the due course.

Experimental Section

General Protocol for Library Synthesis

The following procedure represents the synthesis of a representative set of 96 compounds (plate 4). To a solution of quinolone 5 (1.0 g, 3.52 mmol) in CH₂Cl₂ (30 ml) was added 2,6-lutidine (1.4 ml, 12.3 mmol), TESOTf (2 ml, 8.8 mmol) and a solution of enone B₄ (1.1 g, 7.04 mmol) in CH₂Cl₂ (10 ml). The resulting solution was stirred for 2 h at room temperature and was treated with TFA (0.54 ml, 7.04 mmol). The reaction was quenched by addition of saturated aqueous solution of NaHCO₃ (15 ml), and extracted with CH₂Cl₂ (3 × 30 ml). The organic layer was dried with anhydrous MgSO₄, concentrated in vacuo and purified by flash chromatography on silica gel (ethyl acetate: hexane = 1:3) to give 1.1 g (70% yield) of the corresponding tricyclic ketone. This product (0.72 g, 1.83 mmol) was dissolved in THF (20 ml) and treated with Pd(PPh₃)₄ (10.5 mg, 0.09 mmol) and morpholine (0.2 ml, 3.66 mmol). The resulting solution was stirred for 30 min at 20 °C, concentrated in vacuo and purified by flash chromatography on silica gel (ethyl acetate: hexane = 1:1) to give 580 mg (90% yield) of the corresponding tricyclic amine C (compound 50, see Supporting Information). ¹H NMR (500 MHz, CDCl₃) δ 0.76 (m, 1H), 0.88 (m, 1H), 1.14 (m, 1H), 1.30 (m, 2H), 1.49 (m, 1H), 1.72 (m, 3H), 1.82 (m, 1H), 1.94 (m, 1H), 2.26 (m, 1H), 2.37 (m, 1H), 2.56 (m, 2H), 2.81 (t, 1H, J = 13.4 Hz), 3.44 (br s, 1H), 3.83 (m, 1H), 4.23 (br s, 2H), 4.83 (br s, 1H), 6.68 (d, 1H, J = 8.4 Hz), 7.30 (d, 1H, J = 8.4 Hz), 7.73 (s, 1H); ¹³C NMR (125 MHz, CDCl₃) δ 26.2, 26.4, 26.5, 314, 32.3, 39.0, 41.7, 43.66, 43.74, 46.0, 54.4, 54.9, 116.7, 118.4, 125.1, 127.3, 135.8, 147.9, 194.2, 208.9; MS (APCI) calculated for C₂₀H₂₄N₄O₂ 352.2, found: 387.1 [M+Cl]⁻.

A solution of this amine (350 mg, 0.98 mmol) in CH₂Cl₂ (14 ml) was next divided into 7 equal batches, and treated with pyridine (57 μl, 0.7 mmol) and 7 acid halides (D₁-D₇) (0.42 mmol each). The resulting solutions were stirred for 3 h at room temperature, concentrated in vacuo and purified by flash chromatography on silica gel (ethyl acetate: hexane = 1:2) to give seven corresponding amides E (51-63 mg, 85-99% yield). The parent N-alloc dihydroquinolone was used as the eighth amide. Each of the amides E was diluted in CHCl₃ to a final concentration of 0.2 M, divided into 12 equal batches (20 μl, 4 μmol per well) and arrayed into a polypropylene 96-well PCR plate. The plate was treated with 12 amines F₁-F₁₂ (0.74-1.15 μl) according to the plate map shown in Supporting Information. Each reaction mixture was treated with the 1.3 M solution of NaBH(OEh)₃ in CH₂Cl₂ (12.3 μl per well). Upon completion, the reaction mixtures were transferred onto preparative TLC plates as circular spots using a multichannel pipettor. The plates were developed using ether: hexanes = 2:1. The products were detected under UV light and removed from TLC plates as silica gel pellets, from which the final compounds were eluted with 0.8 ml of CH₃OH. Following analysis of purity of each compound G by TLC, the solvent was removed in vacuo. Twelve randomly selected compounds were dissolved in CD₃OD (0.5 ml) and analyzed by ¹H NMR. The amount of material in each sample was determined by integration using residual CH₃OH as a pre-calibrated internal standard. This protocol was used next to prepare all the remaining library members.

Representative library member 21, which was obtained in 90% yield. ¹H NMR (500 MHz, CDCl₃) δ 0.73 (m, 1H), 0.86 (m, 1H), 1.13 (d, 1H, J = 12.7 Hz), 1.29 (m, 3H), 1.39 (m, 1H), 1.70 (m, 3H), 1.82 (m, 4H), 1.95 (m, 1H), 2.07 (m, 2H), 3.16 (s, 1H), 3.34 (s, 1H), 3.72 (m, 1H), 3.81 (m, 1H), 3.89 (s, 3H), 3.90 (s, 3H), 4.34 (m, 2H), 5.30 (m, 1H), 6.51 (m, 1H), 6.82 (d, 1H, J = 8.0 Hz), 6.89 (m, 2H), 7.11 (d, 1H, J = 8.4 Hz), 7.16 (d, 1H, J = 3.3 Hz), 7.33 (m, 2H), 7.94 (d, 1H, J = 2.0 Hz); ¹³C NMR (125 MHz, CDCl₃) δ 26.4, 26.5, 26.8, 29.8, 30.9, 32.7, 38.7, 40.3, 49.7, 51.5, 52.1, 54.1, 56.0, 111.2, 111.6, 112.2, 118.3, 120.3, 124.3, 125.0, 126.6, 131.6, 133.3, 141.4, 144.8, 147.9, 148.2, 149.1, 160.0, 195.6; HRMS (ESI) calculated for C₃₄H₄₀N₅O₅ [M+H]⁺: 598.3029; Found: 598.3029.