Solid-Phase Synthesis and Combinatorial Technologies

TABLE 9.7 Ligands from Solution-Phase Discrete Monoand Diphosphine Library L24a for the Catalyzed Heck Reaction of Labeled Substrate 9.92 with Supported 9.93: Screening Results

aF1 = no fluorescence, F3 = strong fluorescence.

cases (libraries L26–L28) a preliminary flash chromatography was necessary to separate the library members. The reduction of the mixtures was compared with the reduction of isolated ketones, and an almost complete reproducibility of the results obtained in the two protocols was observed (Table 9.8). Thus, the reduction of ketone mixtures allowed determination of the chiral preference (S or R) and the enantiomeric excess produced by the catalyst in the presence of substrates with different electronic or steric properties.

The rapid assessment of the catalytic profiles of several catalytic systems, tested on substrate libraries, should become a useful tool to speed the selection of optimally catalyzed reaction conditions, providing that faster detection methods affording reliable results in shorter times will be available. Eventually, the parallel use of substrate and catalytic system libraries should become routine to thoroughly assess the catalytic properties and the specificity of a chemical transformation.

9.4.6 An Example: Synthesis and Screening of a Catalytic System Library as Alkene Epoxidation Catalysts

Francis and Jacobsen (174) reported a 5760-member pool library of ligand–metal complexes L30, which is shown in Fig. 9.41 together with its synthetic scheme. First, the monomer set M1 (five α-amino acids, Fig. 9.41) was coupled onto five portions of aminomethyl PS resin using typical peptide coupling conditions to give, after Fmoc

aAverage enantiomeric excess from different libraries (entries 1–10) or from a single library (entries 11–14). bEnantiomeric excess from discrete ketone reduction.

deprotection, compounds 9.113 (steps a and b). After pooling of the resin and splitting into three aliquots (step c), the resin was coupled with the monomer set M2 (three compounds, including a β-aminoalcohol M2,1, an α,β-diamine M2,2, and an α-amino acid M2,3, steps d1–d3, Fig. 9.41) and deprotected (steps e1–e3) to give resin-bound amines 9.114a–o. A structure similar to known epoxidation catalysts was also added, starting from Wang PS resin and coupling it sequentially with an aldehyde to give 9.115 and with building block M2,2 to give 9.114p (Fig. 9.41) as a positive control for screening. The 16 resin aliquots were pooled, split into 12 portions (step f), and coupled with monomer set M3 (12 capping agents, including aldehydes, acids, and a skip codon, steps g1–g2, Fig. 9.41) to give resin-bound imines 9.116 and amides 9.117. The resin aliquots were then pooled (step h) and exposed to monomer set M4 (30 metal ion sources where the metal, the oxidation state, or the counterion was changed) in solution for 1 h (step i) to produce L30 as 30 pools of 192 compounds after filtration and rinsing of the beads (Fig. 9.41). The quality of the library was checked using both visual detection of the expected colors of the ligand–metal complexes and qualitative detection of the complexes using inorganic staining reagents. Approximately 80% of the expected library individuals were detected, and the library was progressed to screening.

The selected reaction was the epoxidation of trans-β-methylstyrene (9.118, Fig. 9.42). At first, various potential reaction conditions were tested for the production of epoxide 9.119 as an enantiomeric couple (chiral GC detection). Among them, hydrogen peroxide and tert-butylhydroperoxide gave good results. The use of the former reagent

476 APPLICATIONS OF SYNTHETIC LIBRARIES

d3: as step a; e1: as step b; e2: absent; e3: as step b.

30 pools of 192 compounds

f: pool; g: resin portioning (1 to 12); h1: DIPEA, DMF/TMOF 9/1, rt; h2: as step a; i: THF/MeOH 4/1, rt.

Figure 9.41 Catalyzed alkene epoxidation: synthesis and structures of the SP pool peptidomimetic catalytic system library L30 and of the monomer sets M1–M4.

a: catalyst, DCM/tBuOH 1/1, rt.

tested reagents: O2, NaIO4, 4-PPNO, NMO, tBuOOH, H2O2

selected: 30% aq. H2O2

Figure 9.42 Catalyzed epoxidation of 9.118: selection of the oxidant.

was described in Ref. 174, while the other is the subject of another report. Deconvolution of the library L30 in the hydrogen peroxide epoxidation of 9.118 was then started.

The 30 L30 pools, varying the M4 constituent, were first screened in the epoxidation reaction (Fig. 9.43). As a control, 30 parallel reactions containing only the metal source were also performed to determine the metal–ligand-dependent catalysis. The results, expressed as relative yields to an internal standard (from 0 = no reaction to 3 = maximum observed), are reported in Table 9.9. Seventeen of the 30 pools showed either ligand-dependent (7) or ligand-independent (10) catalytic epoxidation activity. The most significant ligand-dependent catalytic activity was displayed by Fe ions, with Fe2+ ion and Cl– counterion as the best combination (compare entries 4–6, Table 9.9).

Having determined the best M4 component as M4,7, library L31 was prepared as 12 sublibraries of 16 compounds containing different capping monomers M3 and subsequently exposed to a solution of FeCl2 (Fig. 9.43). The screening results are presented in Table 9.10. Two pyridine-based capping monomers (M3,1 and M3,8) were roughly equivalent to each other and significantly more active than the other 16-member pools in promoting the epoxidation reaction.

The synthesis of the 32 discretes composing the most active pools mentioned above should have produced the most active catalytic system. The authors, though, prepared L32 as 192 discretes, corresponding to all the pools tested above in L31, using the radiofrequency encoding approach with directed sorting (see Sections 7.4.3 and 7.4.4). The identification of three active ligands 9.120–9.122 (Fig. 9.43) followed this third iterative screening. The deconvolutive approach was validated by these results, where all the best ligands contained the expected monomers M3,1 (9.120) and M3,8 (9.121 and 9.122), identified in the previous iterative round. The best monomer in position M2 was also clearly selected as L-serine, while position M1 was represented by two similar monomers, L-serine (9.120 and 9.121) and L-cysteine (9.122).

The three selected ligands were excellent in promoting the epoxidation, but the stereoselection was negligible, varying from 4 to 7%. An additional 96-member parallel library L33 was built to improve the stereochemical outcome of the reaction, building on the SAR acquired from L30–L32. Its structure and the composition of the monomer sets M1–M3 are reported in Fig. 9.44. Three best ligands 9.123–9.125 (Fig. 9.44) were selected using the same screening protocol on L33 as described previously.

a:screening for catalytic activity on epoxidation reaction; b: resin portioning (1 to 12);

c:reaction with M3 and deprotection; d: THF/MeOH 4/1, rt.

HS N

FeCl2

OH N

9.122

Figure 9.43 Deconvolution of the SP pool peptidomimetic library of catalytic systems L30 through lower complexity libraries L31 and L32 and identification of catalytic composites 9.120–9.122 obtained from its screening.

A clean conversion of 9.118 to 9.119 was obtained, with moderate stereoselection (15–20% enantiomeric excess e.e.) for the R,R (9.123, 9.124) or the S,S enantiomer (9.125). Resynthesis of the unbound analogues of 9.120–9.125 was planned to test their effectiveness in the same reaction and to evaluate the influence of the solid support on the reaction outcome. The deconvolution-prone, pooled format for catalyst discovery has been used by Venton (175) for the synthesis of a >28,000-member β-cyclodextrin library (13 pools) as a source of Zn-based catalytic systems with phosphatase-like activity.

TABLE 9.9 Screening of the SP Pool Peptidomimetic Encoded Catalytic Library L30 for Epoxidation of 9.118

aCalculation based on the amount of 9.119 compared with an internal standard.

TABLE 9.10 Activity Screening of the SP Encoded Pool Peptidomimetic Catalytic System Library L31: Deconvolution of M3

aCalculated as the yield of 9.119 compared to an internal standard.

480 APPLICATIONS OF SYNTHETIC LIBRARIES

L33

9.125

Figure 9.44 Structure of the SP peptidomimetic focused catalytic library L33 and of the enantioselective catalytic composites 9.123–9.125 obtained from its screening.

9.4.7 An Example: Synthesis and Screening of an Encoded Acylation Catalytic Library

Taylor and Morken (176) reported the synthesis and on-bead screening of a 3150-mem- ber encoded SP library of potential acylation catalysts L34, whose synthesis and structure are reported in Fig. 9.45. Resin-supported bromide 9.126, obtained by treatment of amino Tentagel resin with bromoacetic acid (step a), was reacted (step b) with the monomer set M1 (primary amines, 15 representatives, Fig. 9.46); then the secondary amine function of intermediates 9.127 was coupled (step c) with monomer set M2 (15 representatives, N-protected α-amino acids, carboxylic acids, and a skip codon, Fig. 9.46). Deprotection of the nitrogen (step d) and capping (step e) with monomer set M3 (carboxylic acids, 15 representatives, and a skip codon, Fig. 9.46) produced L34 as 15 pools, each made of 210 compounds (Fig. 9.45). Each monomer representative was encoded using a popular encoding method (177). The library structure captured the concept of a basic center and a nucleophilic center mutually arranged in different orientations to detect a suitable bifunctional acylation catalyst. A known acylation catalyst (supported monomer M2,12) was added among the monomer set M2 to validate the applicability of the screening and its ability to detect catalysts in an encoded SP pool format.