Thermographic Selection of Effective Catalysts from an Encoded Polymer-Bound Library

See allHide authors and affiliations

Science  10 Apr 1998:
Vol. 280, Issue 5361, pp. 267-270
DOI: 10.1126/science.280.5361.267


A general method is introduced for the rapid and simultaneous evaluation of each member of large encoded catalyst libraries for the ability to catalyze a reaction in solution. The procedure was used to select active catalysts from a library of potential polymer-bound multifunctional catalysts. From ∼7000 beads screened (3150 distinct catalysts), 23 beads were selected for catalysis of an acylation reaction. Kinetic experiments indicate that the most strongly selected beads are also the most efficient catalysts.

In the development of new catalysts, many iterations of design and redesign are usually required to increase catalyst activity. Accordingly, many research groups have begun to use combinatorial chemistry (1) and solid-phase synthesis (2) to rapidly produce large numbers of potential catalysts. Despite recent progress in evaluating the thermodynamics of equilibrium processes involving polymer-bound libraries (3), methods for assessing the kinetics of reactions involving polymer-bound reagents have not been available. This circumstance has prevented the analysis of very large libraries (104 to 106 members), because screening for small-molecule catalysts requires an individual assay for each member of a catalyst library. Here, we report the development of a general single-pot assay for large, encoded polymer bead–bound catalyst libraries, which we used to select active catalysts from ∼7000 encoded catalyst beads (3150 different catalysts) prepared through “split and pool” solid-phase synthesis. This method is directly applicable to libraries of larger size.

Most chemical reactions have a measurable heat of reaction ΔH r o, and thus temperature Thas been used to survey the progress of catalytic reactions (4). Because all catalysts in a parallel library assay are evaluated under the same reaction conditions, the most active catalyst will cause the largest temperature change (ΔT ∝ turnover frequency × ΔH r o). Recent advances in two-dimensional real-time infrared (IR) thermography have made spatial resolution of temperature possible through measurement of blackbody radiation. Therefore, IR thermography may be used to simultaneously compare the rate of each reaction, in an array of reactions, by measuring the relative reaction temperatures. This concept was imaginatively used by Moates et al. (5) for the parallel evaluation of the ignition temperatures of 16 spatially addressed metal-doped alumina pellets in the presence of H2 and O2 gases at elevated temperature. To date, these approaches have not been realized for the analysis of large (104 to 106 members) polymer bead–bound libraries for catalysis of typical solution-phase chemical reactions. Whereas 90°C temperature changes on 3 mm by 4 mm solid pellets were observed in the above heterogeneous assay, much smaller temperature changes were expected in solution because of the effective cooling of the 300- to 500-μm polymer beads by reaction solvent. Absorbance of the IR signal by reaction solvent or reagents or both can also pose a complication. Finally, with polymer-bound catalysts, diffusion of reagents through the polymer matrix might limit the reaction rate and result in a leveling effect such that it would not be possible to differentiate catalyst activities over a certain threshold value.

Our preliminary studies centered around the acyl transfer reaction because, until recently, chemists have met with little success in approaching the selectivity of enzymatic acylation catalysts found in nature (6). On the basis of nature's example, multifunctional catalysts that use a concerted interplay between tethered functional groups hold great promise as particularly potent effectors of chemical transformations. The choice of functional groups as well as their relative position and orientation in space are all critical factors for effective rate acceleration. According to proposed mechanisms for various nucleophile catalyzed acylation reactions (Fig.1) (7), we reasoned that a suitable base covalently tethered in the correct orientation relative to a nucleophilic center might increase catalyst activity through a bifunctional catalytic manifold. Additional interactions, for example, those that stabilize the intermediate acylpyridinium salt or that bind to the reacting alcohol, might lead to further increases in reactivity.

Figure 1

Mechanism for the nucleophile (Nu) catalyzed reaction between acetic anhydride and alcohols. Ac, acetyl; Me, methyl.

To develop a thermographic assay for our envisioned library, we used an IR camera (Cincinnati Electronics IRRIS 256ST, 256 × 256 InSb FPA detector) to examine both noncatalyst beads (acylated 300-μm tentagel S-NH2; Rapp Polymere) and those with a known acylation catalyst attached (N-4-pyridylproline coupled to polymer resin) (8). Although in the reaction solution (8:1:1:1 chloroform:ethanol:acetic anhydride:triethylamine) it was not possible to see individual noncatalyst beads with the IR camera (bead temperature rapidly equilibrates with solvent), catalyst beads exhibited a sustained ∼1°C temperature increase from that of the bulk solvent. This temperature difference is easy to observe with the IR camera (detection limit = 0.02°C; see Fig. 2A), indicating that active catalyst beads can be reliably distinguished from those that are inactive. With chloroform as solvent, the beads rose to the top of the reaction solution, thus avoiding solvent interference with IR transmission. When the proportion of chloroform was reduced such that the beads sank, it was not possible to observe hot beads with the camera.

Figure 2

(A) Infrared thermographic image of ∼20 catalyst beads in the presence of ∼3000 noncatalyst beads. Arrows indicate 2 of the 14 visible hot beads. (B) Closeup IR thermographic image of the trimeric catalyst library in the presence of acylation reagents, showing one hot bead being selected for decoding (tweezers in upper left).

With a working assay developed, an encoded (9) library was prepared by split and pool solid-phase synthesis (10), with the reaction sequence and monomers shown in Fig.3 (11). In addition to a variety of basic groups and a diverse collection of potentially nucleophilic compounds, a subset of library monomers was chosen at random on the chance that they might act through an unpredicted mode of catalysis. Additionally, N-4-pyridylproline (B12) was included as a control in the second monomer group to ensure that effective catalysts were present. For library synthesis, initial displacement of an activated bromide with a variety of primary amines (A1 to A15) was followed by coupling a variety of protected amino acids (B1 to B15) to the resulting secondary amine nitrogen. After deprotection, a collection of carboxylic acids (C1 to C15) was coupled to the liberated amine terminus, thereby completing the trimeric library. With 15 monomers in each position (including a skip codon) (12), the library should be composed of 3150 distinct compounds.

Figure 3

Synthesis scheme and monomers used in the three positions of the trimeric catalyst library. (A) Amine monomers, (B) amino acid monomers, (C) acid monomers, and (D) library synthesis. Boc, butoxycarbonyl; Bn, benzyl.

The addition of 610 mg of library resin beads (∼7000 beads) to a solution composed of 40 ml of chloroform, 6 ml of ethanol, 6 ml of triethylamine, and 3 ml of acetic anhydride revealed beads with a range of temperatures; however, the bulk maintained temperatures close to background. Appropriate adjustment of the IR camera's temperature display span allowed for visualization of only the hottest beads. Over time, average bead temperature appeared to decrease, presumably because of the consumption of reagents; the addition of fresh reagents increased bead temperature. Of the hottest beads, 23 were selected (Fig. 2B) and subsequently decoded (not all of the hottest beads were selected for decoding). As shown in Table1, of the 23 selected and decoded beads, 21 were either 1 or 2, prepared from aminesA12 and A13, coupled to acidN-4-pyridylproline (B12). Attachment of monomerB12 in the amino acid position effectively terminates compound synthesis because B12 does not have an amine on which to couple an acid in the third position. In addition to1 and 2, beads containing sequences coding for compounds 3 and 4 were also selected.

Table 1

Selection frequency and structures of beads from the IR catalyst library assay. Ph, phenyl.

View this table:

To ascertain whether the assay reflects catalyst activity, we studied compounds of interest with a series of kinetic experiments. After resynthesis on aminomethyl polystyrene macrobeads, five beads of a given catalyst were added to 1.1 ml of an 8:1:1:1 solution of chloroform:acetic anhydride:1-butanol:triethylamine. The average percentage of reaction at 9 min for three kinetic runs is shown in Fig. 4. After subtraction of background reaction (13% at 9 min), noncatalyst beads (N-acetate–capped beads, 6) showed little catalytic activity (∼1% conversion after 9 min). Hot beads1, 2, and 3 all gave substantial conversion (39, 24, and 23%, respectively), as compared with the noncatalyst beads and structure 5 (9% conversion), a compound that should have been present in the library but was not selected. Beads containing compound 4 gave no rate acceleration above background, indicating that the wrong bead may have been selected during the assay. When B12 was attached directly to beads and used in these kinetic runs, 14% conversion was achieved at 9 min. Thus, A12 appears to enhance the catalytic activity of the N-4-pyridylproline nucleus. Although it is peculiar that both A12 and its enantiomerA13 were selected from the assay whereas nonracemicl-proline was used for the synthesis ofB12 (13), control experiments indicated that, under slow amide-coupling conditions, B12 is racemized and similar enantiomeric diastereomer mixtures are likely present on both beads 1 and 2. At this point, no mechanistic conclusions can be drawn in regard to the enhanced activity of1 relative to polymer-bound N-4-pyridylproline.

Figure 4

Percentage of conversion above background (13%) at 9 min for polymer bead–bound compounds. The conditions were as follows: five beads (500 μm polystyrene, 1.04 mmol/g) of a given compound were added to a stirred solution of 1.1 ml of 8:1:1:1 CHCl3:triethylamine:n-BuOH:Ac2O. Conversion was measured by gas chromatography versus zero time sample with an internal standard.

Several points in regard to the assay merit mention. Structure1, one of the most strongly selected beads (11 beads selected from the ∼30 present in the assay), also showed the most efficient catalysis in the kinetic runs as compared with all others. This result indicates that, to a rough approximation, selection frequency reflects catalyst activity and may be used as an indicator of catalyst efficiency (14). Statistically, 1 bead in 15 should have B12 in the second monomer position, and compounds1 and 2 should each be present once in every 225 beads. Because B12 is a cap, compounds containing this monomer are present at a frequency 15 times greater than that of any other compound. In the entire 7000 beads screened, ∼60 beads should be either compound 1 or 2. That it was possible to select 21 of these beads to the near exclusion of the other ∼400B12-containing beads present highlights the reliability of the assay. Also of note is the extent to which the assay can discriminate between similar levels of activity: Compound 5, a catalyst with about fourfold less activity than 1, was present in the library assay (∼30 beads) but was not selected.

Although reactions with lower catalyst turnover rates might be more challenging, evaluation of organometallic libraries or libraries of ligands for ligand-accelerated metal catalysis should not require substantial modifications to the above described protocol. Efficient catalysis of the acylation reaction studied here is reported to be exothermic (ΔH r o = −14.9 kcal/mol) (15); however, it does not involve an atypical enthalpy change, as compared with many reactions of current interest in both academic and industrial synthesis (Table2). In general, analysis of combinatorial libraries should reveal novel catalyst structures with potentially new modes of catalysis, as observed here.

Table 2

Calculated enthalpy change (ΔH r o) for selected chemical reactions. ΔH r o was calculated from thermochemical data reported in (15). t, tert; Bu, butyl.

View this table:
  • * To whom correspondence should be addressed. E-mail: morken{at}


View Abstract

Stay Connected to Science

Navigate This Article