Solid-Phase Synthesis and Combinatorial Technologies

S

Figure 8.11 Synthetic scheme to the solution-phase, discrete thiohydantoin library L2.

is covered in the following sections, while automated work-up and purification procedures are covered in Sections 8.3–8.5.

8.2.4 Automation and Solution-Phase Discrete Libraries

The synthesis of small solution-phase discrete libraries does not require automation, and a set of multichannel pipettes is enough to deliver solvents and reagents into the reaction vessels. The reaction vessels are arranged in racks that spatially isolate each vessel and allow the transfer of the whole array onto an orbital stirring unit, onto a heating/cooling instrument, or into a closed system under an inert atmosphere. Even several hundred discrete reactions can be performed and processed rapidly by manual techniques. Some commercially available instruments are able to handle in parallel a few tens of compounds (stirring, heating, cooling) (58, 59), while others also automate the work-up/purification procedures allowing the manual delivery of reagents even under an inert atmosphere (60, 61).

Robotic dispensing units based on pipetting arms can be used to partially automate library synthesis in solution, which included purification steps consisting of liquid– liquid or solid–liquid extraction followed by withdrawal of the desired phase or by elution of the product from the SP (62, 63; see Section 8.3). The dispensing units of automated SP synthesizers can be used for this purpose, and sometimes manufacturers claim that it is possible to run solution-phase library synthesis on these automated instruments, even though such claims must be validated for many operations. Dispensing units are the core of fully automated modular systems specifically designed for solution-phase synthesis (64, 65). They contain stirring/vortexing and heating/cooling

354 SYNTHETIC ORGANIC LIBRARIES: SOLUTION-PHASE LIBRARIES

O

RO R1

N

N

O

O

18 monomers

X

X = H, 2-MeO, 4-MeO, 4-NO2, 2-F, 4-F, 2-Br, 4-Br, 2-Me, 4-Me

AcO

O

AcO

AcO

OAc

19 monomers

Figure 8.12 Synthesis of the solution-phase, discrete thiohydantoin library L2: structure of the monomer sets M1–M3.

8.2 SOLUTION-PHASE DISCRETE LIBRARIES 355

modules and allow the fully automated, parallel handling of several arrays processed through different modules at a given time. Many of the most commonly encountered organic reaction conditions, including working under an inert atmosphere (via rubber septa that are pierced by robotically controlled needles), are compatible with these instruments. The example described in the next section is based on the use of a fully automated synthesizer to generate a small solution library. A recent report (66) described the use of several synthesizers for different operations in the same automated solution library synthesis, optimizing the performance of each step and reducing the lag times.

Work up/purification procedures have also largely benefited from the commercial availability of semiautomated or automated instrumentation. The panel of tools spans from small, 96-well based devices to increase the throughput of manual operations (67, 68) to semiautomated systems able to purify in parallel combinatorial samples (69) or to concentrate in parallel large, discrete solution libraries (70, 71). Several systems developed in-house have also been recently reported (72, 73), and a recent review covered the most recent trends in automated high-throughput purification methods for solution discrete libraries (74).

Some companies (75, 76) have developed their own automated instruments for combinatorial library synthesis in solution to produce large, purified arrays of discretes (up to several tens of thousands of individuals); the available information is obviously scarce, but an example of such a proprietary integrated synthesizer will be presented in section 8.2.6.

8.2.5 An Example: Synthesis of a Triazine Library

Whitten et al. (77) reported the synthesis of a focused triazine library L3 (Fig. 8.13) of >350 members based on the known corticotrophin releasing factor 1 (CRF-1) antagonist, 8.24 (Fig. 8.13), to discover more potent triazine analogues. The two-step synthetic scheme leading to L3 from dichlorotriazines 8.25a,b using two amine monomer sets M1 and M2, is reported in Fig. 8.13.

The first chlorine displacement was carried out at rt, while the second required heating. Both reactions are well known in the literature and chemical assessment was not considered necessary. A modular automated system (64) allowed the simultaneous processing of several arrays of glass vials (1.8 mL each, typically 5–25 vials per array, each stoppered with a rubber septum) through the stirring unit, the heating unit, and the extraction/purification modules. The results of the first arrays were used as both rehearsal of the monomers and a model library synthesis. Most of the amine monomers were commercially available, while the primary amines 8.27 were also prepared on the automated instrument (Fig. 8.14). The library synthesis was quite successful, with around 70% library individuals prepared in milligram amounts and successfully QCed by HPLC/MS (purity >70%). The confirmed compounds were tested as CRF-1 antagonists, while the other samples were discarded.

The only manual operations required for the synthesis of this library were the preparation of stock solutions of 8.25a,b (0.5 M THF) and the monomer representatives M1 (1.5 M DIPEA and 3 M THF) and M2 (1 M THF). The other operations

356 SYNTHETIC ORGANIC LIBRARIES: SOLUTION-PHASE LIBRARIES

N

N N

NN H

8.24

discrete library

a: THF, rt, 10'; b: THF, 65°C, 1 hr.

M1 : substituted anilines

M2 : primary and secondary alkyl amines

Figure 8.13 Synthesis of the solution-phase, discrete triazine libraryL3 inspired by the CRF-1 antagonist 8.24.

were controlled and automated by the instrument software. A brief description of the main software commands is reported in Fig. 8.15 (Method 1):

•2STEP1/WAIT2 implied the addition of 8.25a or 8.25b (0.04 mL, stock solution) to septum-sealed 1.8-mL vials followed by the monomer set M1 (0.026 mL, stock solution). The reactions were left standing for 10 min at room temperature (WAIT2).

a: dry Et2O, rt; b: BH3, THF, rt.

Figure 8.14 Synthesis of a cyclopropane-based monomer subset for the solution-phase, discrete triazine library L3.

Figure 8.15 Software automated protocols for the synthesis of the solution-phase, discrete triazine library L3.

•2STEP2/WAIT2 treated the resulting solutions of 8.26 with 10 to 20 equivalents of M2 (stock solutions) and heated them at 65 °C for 1 h. The reactions were then cooled to rt (WAIT2).

•2STEP3 dispensed EtOAc (0.5 mL) and 1 N HCl (0.5 mL) to each vial and transferred the organic layers to new vials after partition.

•ROBOTFAS analyzed all the vials by GC-MS, then evaporated their contents to dryness with nitrogen, producing crude L3 as discretes with good average yields and purities.

The synthesis of 8.27 (Method 2) consisted of the following:

•CYCLOPTR, where the acyl chloride and the amines in THF were stirred for 1 h before partitioning the crude between ether and 1 N HCl. The organic phases were transferred to new vials and washed with aqueous sodium hydrogen carbonate and then to a second vial for drying with sodium sulfate. The ethereal solutions were transferred to vials containing borane and stirred for 6 h, after which time the reactions were quenched with aqueous sodium hydrogen carbonate. The organic layers were transferred to new vials and heated at 125 °C for 9 h.

•ROBOTFAS was performed as in the above sequence.

Fully automated instruments with low–medium throughput, typically a few hundred compounds per week for oneto three-step synthetic schemes in solutions, are suitable for small, focused libraries where more challenging reaction conditions such as heating, using reactive intermediates, or when inert atmospheres are needed. These instruments are also suitable for intermediate steps in the construction of larger libraries. For example, monomer rehearsal can be performed by reacting a common intermediate with various monomer candidates, and a small model library may be prepared. Even chemistry assessment may be tackled with these instruments.

Among some recent papers, the same researchers (78) reported the synthesis of two libraries of piperazines and piperidines containing 1086 and 835 discretes, respectively, via one-step alkylation or acylation of several scaffolds. Bhat et al. (79) used the same instrument to prepare a 26-member library of paclitaxel C7 esters via a three-step procedure. An array of twenty 1,2,4-oxadiazoles (80) was prepared through

358 SYNTHETIC ORGANIC LIBRARIES: SOLUTION-PHASE LIBRARIES

a three-step procedure from carboxylic acids and benzamidoximes using a semiautomated device (60). Frank et al. (81) reported a 107-member quinolone library prepared through a five-step protocol with a semiautomated device developed in-house.

8.2.6 An Example: Synthesis of a Spiro[Pyrrolidine-2,3′-Oxindole] Library

Powers et al. (82) reported the synthesis of a 1280-member library of discrete chalcones in solution (L4, Fig. 8.16) with an extremely high (96% average) compound purity. This library was used as such or through subsets to generate several primary libraries as sources of novel biologically active compounds. A 25,600-member library

R2 L5

25,600 discretes spiropyrrolidine library

Figure 8.16 Synthesis of the solution-phase, discrete spiropyrrolidine libraryL5 from a subset of the solution-phase chalcone library L5.

8.2 SOLUTION-PHASE DISCRETE LIBRARIES 359

of spirotricyclic compounds (L5, Fig. 8.16) that was derived from the chalcone monomer subset M1 (80 representatives) and from full monomer sets M2 and M3 (16 isatins and 20 α-amino acids, respectively) following the synthetic scheme reported in Fig. 8.16 was described by Fokas et al. (83).

The feasibility of the proposed chemical route was assessed, and a brief monomer rehearsal was carried out by reacting the unsubstituted chalcone 8.28 with several isatins (8.29a–d, Fig. 8.17) and α-amino acids (8.30a–g, Fig. 8.17). Most of the single reactions produced the desired spiro compounds in good yields and purities and also with excellent regioand stereocontrol. The adverse steric interaction between the chalcone carbonyl and the R4 substituent gave control over the regiochemistry and only one diastereomer (as a pair of enantiomers) of 8.31a–j was generally obtained (Fig. 8.17). Two monomer candidates M3, 8.30f,g, performed poorly and were discarded. The yields and purities obtained during this phase of study are reported in Table 8.1.

The large majority of the selected monomer representatives were commercially available and a few were prepared in a single reaction. The library synthesis was performed on a 50 µM scale (around 20 mg/compound) using a 96-well-based, proprietary automated and integrated instrumentation called the Automated Molecular Assembly Plant (84), which is suitable for the production of medium–large libraries

R2

8.31a-j (enantiomeric mixtures)

Figure 8.17 Synthesis of the solution-phase, discrete spiropyrrolidine library L5: monomer rehearsal.

360 SYNTHETIC ORGANIC LIBRARIES: SOLUTION-PHASE LIBRARIES

of discretes in solution. Twenty-five percent of the resulting racemic mixtures of compounds were characterized by HPLC/MS, and generally, good purity was observed, with the exception of three amino acids and a single chalcone that gave unsatisfactory results; the corresponding products were discarded. A more extensive rehearsal of the monomers could have prevented this undesired outcome, which led to the loss of (3 × 1280) + (1 × 272) = 4112 (16% of the total library population) expected individuals from the library.

Several chalcone-based libraries (see Fig. 4.11, section 4.2.2) were combined to give a total of over 74,000 discretes (82) and, while no mention of time and resources required was made in the paper, the significant effort required to develop the automated equipment for the synthesis of medium–large libraries in solution in-house was surely paid back by the potential to continuously produce primary lead-seeking libraries in solution in large quantities. Only a few specialized companies may invest so heavily in combinatorial automation, but any interested party can access these companies to contract a library synthesis starting from a proprietary chemistry of interest.

The same group (85) has also reported the synthesis of a > 40,000-member triazine library from multiple decoration of cyanuric chloride. Another group that developed its own automated instrumentation reported the synthesis of a 1920-member tetrahydroquinoline library using a three-component cycloaddition between anilines, aldehydes, and alkenes (86).

TABLE 8.1 Spiro [pyrrolidine-2,3-oxindole] Discrete Library L5: Chemistry Assessment and Monomer Rehearsal

aIsolated compounds. bBy HPLC.

cR3R4 = (CH2)3, proline. dR3R4 = CH2SCH2, thiaproline.

eR3R4 = (CH2)4, pipecolinic acid. fR4R5 = cyclohexyl.

8.3 PURIFICATION OF SOLUTION-PHASE LIBRARY INTERMEDIATES AND FINAL COMPOUNDS: LIQUID–LIQUID AND SOLID-PHASE EXTRACTION SYSTEMS

8.3.1 General Considerations

Organic chemistry in solution has always relied on the production of compounds in high yields, and much effort has been spent in creating, refining and applying a wide set of chemical transformations to a variety of synthetic problems. The available reactions are largely robust enough to successfully plan and execute a given chemical synthesis. Once the desired product is obtained, though, it must be purified from the reaction by-products, and typically this is done via a work-up phase where common separation techniques (extraction, evaporation, precipitation, filtration) produce a crude that is submitted to purification that is usually based on a chromatographic separation.

The switch from single compounds to a combinatorial library in solution increases the complexity of the potential issues to be addressed, as we have seen in the previous chapters. The same is true for the purification of these libraries, which cannot rely on simple filtration and washing of the resin beads, as in the SP chemistries. The separation techniques of classical chemistry are used. However, general, automated methods applicable to all members of a library have to be found. Usually, these methods are also applicable to the final purification of cleaved SP libraries, either as discretes or as pools.

While evaporation is used for the concentration and removal of solvents, usually the reaction by-products are not volatile. Similarly, filtration of precipitated or crystallized solids is not likely to be applicable to all the members of a library, and furthermore the automation of these processes is not straightforward; an interesting example of general precipitation of library members from an organic medium due to the presence of a basic ionizable group has been recently reported by Perrier and Labelle (87). Extraction procedures possess the desired separation properties and have been used for the purification of several solution-phase libraries; we will cover this subject in more depth in this section. An excellent review (88) has recently been published in which the interested reader will find a description of available strategies for separation and purification of single compounds and arrays.

8.3.2 Liquid–Liquid Extractions: Two-Phase Systems

The partition of a reaction between an organic and an aqueous phase is an excellent method to isolate either an organic product from water-soluble impurities or a hydrophilic product from organic impurities. In a reaction the use of water-soluble reagents, catalysts, and coupling agents that are sequestered from the reaction products by simple washing with water is often preferred. An example is 1-(3-dimethylamino- propyl)-3-ethylcarbodiimide hydrochloride (EDC), a water-soluble carbodiimide typically used for amide couplings. More exotic two-phase extraction systems using two immiscible organic solvents, while possible, have not received much attention due to their

362 SYNTHETIC ORGANIC LIBRARIES: SOLUTION-PHASE LIBRARIES

limited applicability to the purification of organic compounds, an exception being the so-called fluorous biphase systems (89), whose principle is covered in detail in Section 8.3.4.

Acidic, neutral, or basic aqueous solutions make it possible to change the phase into which a compound partitions. For example, a basic reaction product is extracted from a water-immiscible organic solvent using an acidic aqueous solution, thus removing organic impurities (step a, Fig. 8.18). Basification of the aqueous phase and extraction with an organic solvent (step b, Fig. 8.18) allows the removal of water-sol- uble impurities and recovery of the basic organic compound in a reasonably pure state. Phase switching can be obtained by tagging neutral organic compounds (step e, Fig.

a:acidic aqueous/organic extraction; b: neutralization, then aqueous/organic extraction; c: discarded;

d:product recovery; e: chemical tagging; f: de-tagging.

Figure 8.18 Liquid–liquid biphase extraction: basic principles applied to basic (top) and neutral products (bottom).