Ruminations regarding the design of small mixtures for biological testing

Synthesis and screening of compound mixtures offer avenues to increase throughput and reduce cycle time in the discovery of new drugs and agrochemicals. Equations are derived which show that the efficiency of synthesis and screening of mixtures is a function of the screening hit rate and the number of compounds in each mixture when simple one-step deconvolution by retesting the individual compounds in each active mixture is employed. Values of hit rate and number of compounds in each mixture which afford various levels of increased efficiency are delineated. Two-step deconvolution, in which the active mixtures from the first round of testing are subdivided into mixtures with fewer compounds for a second round of mixture screening prior to final testing of individual compounds, is shown to be more efficient than simple one-step deconvolution under most conditions. For optimum efficiency, the number of compounds in each mixture in the second round testing should be the square root of the number of compounds in each mixture in the first round. At high hit rates the efficiency of the double scan or indexed approach to deconvolution is shown to be higher than that of simple deconvolution. This discussion is oriented mainly toward mixtures of 4-20 compounds and screens which give hit rates in the 1-10% range. The equations describing efficiency are applied in the context of a 49-member amide library produced as mixtures of seven compounds. This library includes the commercial herbicide pronamide and was screened for herbicidal and insecticidal utility.     

Computer Aided Analysis of Split and Mix Combinatorial Libraries

Combinatorial chemistry using split and pool synthesis involves making and testing mixtures of compounds in pools which are subsets of the larger compound collection. These subsets are created during the synthesis of the collection through a resin splitting and mixing method. Tests are conducted on each of the final pools of mixtures and the individual compounds within a mixture of interest are then identified through some deconvolution scheme, originally involving selective re-synthesis. It is possible that different schemes for splitting and mixing will have different consequences on the overall effort necessary to deconvolute interesting mixtures. The evaluation of different protocols of splitting and mixing involves consideration of more possibilities than can be exhaustively or optimally determined manually in a realistic time frame for most compound collections. We present herein a computational scheme to aid in this analysis. The approach exhaustively examines possible splitting and mixing strategies for the interrelated values of total library size, number of combinatorial steps, number of reaction vessels, and number of compounds per final pool. Weighting factors may be introduced into the various steps. The resulting complete list of splitting and mixing options is scored based on a variable weighting strategy for the total effort of synthesis and deconvolution. The results indicate the splitting/mixing strategy used has an impact on overall efficiency and should be considered in the design of compound libraries.     

Inhibitors of cell migration that inhibit intracellular paxillin/alpha4 binding: a well-documented use of positional scanning libraries

Screening combinatorial libraries for inhibition of Paxillin binding to the cytoplasmic tail of the integrin alpha4 provided the first inhibitors of this protein-protein interaction implicated in enhanced rates of cell migration and chronic inflammation. The preparation of substructure analogs of the lead identified features required for activity, those available for modification, and those that may be removed. The most potent lead structure was shown to inhibit alpha(4)beta(1)-mediated human Jurkat T cell migration in a dose-dependent manner, validating the intracellular Paxillin/alpha4 interaction as a useful and unique target for therapeutic intervention. Moreover, the lead structure emerged from a library that was prepared in two formats: (1) a traditional small mixture format composed of 100 mixtures of 10 compounds and (2) a positional scanning library. Their parallel testing provided the rare opportunity to critically compare two approaches.     

The synthesis and evaluation of a solution phase indexed combinatorial library of non-natural polyenes for reversal of P-glycoprotein mediated multidrug resistance

A combinatorial library of polyenes, based on (-)-stipiamide, has been constructed and evaluated for the discovery of new multidrug resistance reversal agents. A palladium coupling was used to react each individual vinyl iodide with a mixture of the seven acetylenes at near 1:1 stoichiometry. The coupling was also used to react each individual acetylene with the mixture of six vinyl iodides to create 13 pools indexed in two dimensions for a total of 42 compounds. Individual compounds were detected at equimolar concentration. The vinyl iodides, made initially using a crotylborane addition to generate the anti1,2-hydroxylmethyl products, were now made using a more efficient norephedrine propionate boron enolate aldol reaction. The indexed approach, ideally suited for cellular assays that involve membrane-bound targets, allowed for the rapid identification of reversal agents using assays with drug-resistant human breast cancer MCF7-adrR cells. Intersections of potent pools identified new compounds with promising activity. Aryl dimension pools showed R = ph and naphthyl as the most potent. The acetylene dimension had R' = phenylalaninol and alaninol as the most potent. Isolated individual compounds, both active and nonpotent, were assayed to confirm the library results. The most potent new compound was 4ek (R = naphthyl, R' = phenylaninol) at 1.45 microM. Other nonnatural individual naphthyl-amide compounds showed potent MDR reversal including the morpholino-amide 4ej (1.69 microM). Synergistic activities attributed to the two ends of the molecule were also identified. Direct interaction with Pgp was established by ATPase and photoaffinity displacement assays. The results indicate that both ends of the polyene reversal agent are involved in Pgp interaction and can be further modified for increased potency.     

A combinatorial approach toward the discovery of non-peptide, subtype-selective somatostatin receptor ligands

The tetradecapeptide somatostatin is widely distributed throughout the body and is thought to be involved with a variety of regulatory functions. Recently, five human somatostatin receptors (hSSTR1-5) have been cloned and characterized. Several selective peptidal agonists of the hSSTR receptors are known, and we sought to apply this information to the design of novel non-peptide small molecule ligands for each receptor. Initial computational methods identified a 200 nM murine SSTR2 active compound via a database search of our sample collection. A combinatorial library was designed around the structural class of the compound with the goal of rapidly developing this initial lead into the desired subtype-selective small molecules in order to characterize the pharmacology of each of the receptor subtypes. The library was synthesized using the resin-archive, iterative deconvolution format. The total number of unique compounds in the library was expected to be 131,670, present in 79 mixtures of 1330 or 2660 compounds per mixture. Through sequences of screening and mixture deconvolution, the components of selective and highly active (Ki = 50 pM to 200 nM) non-peptide small molecule ligands for somatostatin subtypes 1, 2, 4, and 5 were identified. In addition to discovering compounds with the desired activity and selectivity, useful structure/activity information was generated which can be used in the design of new compounds and second-generation combinatorial libraries.     

Synthesis of a positional scanning library of pentamers of N-alkylglycines assisted by microwave activation and validation via the identification of trypsin inhibitors

A positional scanning library of 625 N-alkylglycine pentamers has been synthesized on solid-phase, employing a set of 10 commercially available primary amines as a source of chemical diversity. The iterative synthetic steps were carried out in tea bags and accelerated by using microwave assisted organic synthesis (MAOS). The reactivity study of the primary amines used as diversity sources led to determine their relative reactivity values and equireactivity factors, which were applied to the library synthesis to ensure comparable concentrations of all final oligomers in the mixtures. This library was validated by the screening, deconvolution, and identification of trypsin inhibitors. These compounds are of potential interest for controlling the intracellular transport of TRPV1 channel.     

Selection of New Sequence‐Selective Unnatural Peptides Binding to Double‐Stranded Deoxyribonucleic Acids (dsDNA) by Means of a Gel‐Retardation Experiment for Library Analysis

Previously, we developed a methodology for the solid‐phase screening of peptide libraries for interaction with double‐stranded deoxyribonucleic acids (dsDNA). In the search for new and more‐potent DNA ligands, we investigated the strategy of solution‐phase screening of chemical libraries consisting of unnatural oligopeptides. After synthesis of the selected amino acid building blocks, libraries were constructed with the general structure Ac‐Arg‐Ual‐Sar‐X¹‐X²‐X³‐Arg‐NH₂, where X represents each of twelve unnatural or natural amino acids. Optimization of the sequence of binding peptides was performed with an iterative deconvolution procedure. Selection of interacting peptides was carried out in solution by means of gel‐retardation experiments, starting with libraries of 144 compounds. A 14‐base‐pair double‐stranded DNA fragment was chosen as the target. After several cycles of synthesis and screening of libraries and individual peptides, an oligopeptide was selected with an apparent dissociation constant of 9⋅10⁻⁵ M , as determined by gel‐retardation experiments. This peptide was studied by NMR spectroscopy. A certain degree of conformational pre‐organization of the peptides was shown by temperature‐dependent circular‐dichroism experiments. Finally, DNase‐I‐footprinting studies indicated a preferential interaction with a 6‐base‐pair mixed sequence 5′‐CTGCAT‐3′. This study demonstrates that gel‐shift experiments can be used for the solution‐phase screening of library mixtures of peptides against dsDNA. In general, this technique allows the selection of new sequence‐selective dsDNA‐interacting molecules. Furthermore, novel dsDNA‐binding unnatural oligopeptides were developed with affinities in the 0.1 mM range.     

Characterisation of combinatorial libraries of mucin-2 antigen peptides by high-resolution mass spectrometry

An epitope motif, TX(1)TX(2)T, of mucin-2 glycoprotein was identified by means of a mucin-2-specific monoclonal antibody, mAb 994, raised against a synthetic mucin-derived 15-mer peptide conjugate. For determination of the epitope sequence recognised with highest affinity by mAb 994, a combinatorial approach was applied using the portioning-mixing technique excluding Cys. Antibody binding of libraries was most profound when Gln was at the X(1) position. Analytical characterisation of the TQTX(2)T library was conducted by amino acid analysis and matrix-assisted laser desorption/ionisation time-of-flight (MALDI-TOF) and electrospray ionisation Fourier transform ion cyclotron resonance (ESI-FTICR) mass spectrometric methods. Control libraries were prepared by mixing 19 individual peptides corresponding to the TQTX(2)T sequence. Thus, mixtures of 6, 10 and 19 pentapeptides were analysed and compared with the combinatorial mixture. MALDI-TOFMS was able to detect only partially the components in the 6- and 10-member mixtures, but failed to characterise a more complex 19-member mixture. In contrast, ESI-FTICRMS resolved all mixtures of higher complexity and provided direct identification at monoisotopic resolution, such as for a peptide library containing 'isobaric' lysine and glutamine (Delta m = 0.0364 Da). The results of this study suggest that ESI-FTICRMS is a powerful tool for characterisation of combinatorial peptide libraries of higher complexity.     

Schiff bases of indoline-2,3-dione (isatin) with potential antiproliferative activity

ABSTRACT: BACKGROUND: Cancer is one of the most dreaded diseases and it is a leading cause of mankind death worldwide. Recent reports documented a remarkable antiproliferative activity of isatin nucleus against various cancer cell lines. The current work describes the antiproliferative activity of Schiff bases of combinatorial mixtures of the isatin derivatives M1-M22 as well as the individual compounds 1-11(A-K) of these combinatorial mixtures. RESULTS: The designed combinatorial library composed from eleven hydrazides A-K and eleven isatin derivatives 1-11 has been synthesized to formally generate 22 mixtures, M1-M22 of 121 Schiff bases, and their antiproliferative activity against K562 chronic myelogenous leukemia cells was evaluated. The indexed method of analysis of the prepared library was applied to elucidate the active components in the tested mixtures M1-M22. The predictions from the crossing procedure was validated through evaluation of the antiproliferative activity of individual compounds 1-11(A-K) of the library. Individual compounds 1-11(A-K) were also evaluated against the non-tumorigenic MCF-12A cell line to investigate their selectivity. A pharmacophore model was developed to further optimize the antiproliferative activity among this series of compounds. CONCLUSIONS: Variable antiproliferative activity was revealed with the investigated mixtures M1-M22 and the individual compounds 1-11(A-K). Most of the tested mixtures and several individual Schiff bases displayed high potency with IC50 values in the low micromolar range. A considerable selectivity of some individual compounds to the tumorigenic K562 cell line compared with the non-tumorigenic MCF-12A cell line was observed as indicated by their selectivity index (SI).     

One-bead-one-compound library of end-capped dipeptides and deconvolution by microflow NMR

As part of our program to identify novel small molecules with interesting biological activity, we have designed and synthesized a library of end-capped dipeptides with an emphasis on compound diversity, complexity, and membrane permeability. An approximately 1500-member library was synthesized manually on large polystyrene beads using the mix-and-split method. The final compounds were cleaved into 384-well plates to generate individual stock solutions for input into high-throughput biological screens. Individual compounds were decoded using a combination of mass spectrometry and microflow NMR spectroscopy. In principle, this approach to deconvolution obviates the need for complicated binary encoding-decoding strategies for one-bead-one-compound libraries.     

Frontal affinity chromatography for the screening of mixtures

A protein stationary phase for frontal affinity chromatography was prepared, containing biotinylated beta-galactosidase immobilized to controlled pore glass beads via covalently bonded streptavidin. Single microaffinity columns of approximately 30 pmol of active beta-galactosidase were prepared from this material and characterized with a known ligand by frontal analysis. These columns were used to measure the specific interactions between the bound beta-galactosidase and a library of modified beta-galactopyranosides using electrospray mass spectrometry as the means of detection. The library contained 89 entries, each representing 4 diastereomers for a total of 356 library members. A single entry was analysed revealing differential activity among the 4 isomers. The library was grouped into 10 mixtures of 24-40 members each with each mixture infused under frontal chromatographic conditions. This deconvolution procedure led to the identification of 34 entries containing isomers with K(d) values better than 10 microM. A method based on a displacement principle was implemented as a rapid prescreen which served as the basis for a parallel column high throughput screening assay.     

Quality assessment of combinatorial pooled libraries using mass spectrometry

Parallel synthesis techniques aim to prepare collections of single compounds which, once tested, can easily be identified by their sole location in the synthesic array. On the other hand, true combinatorial chemistry produces libraries of compounds as mixtures of variable size which require a deconvolution procedure for identification of the active hits or leads. In the latter case, analytical methods are crucial for the success of the strategy and mass spectrometry plays a major role. If the goal is to identify all the library components, including expected products as well as by-products, various mass spectrometric techniques may be necessary. Library components can be separated according to their mass by increasing mass resolution or by their elution time by coupling liquid chromatography and mass spectrometry. The efficiency of such separation techniques are discussed as a function of the size and the degeneracy of the library. Library members possess common structural features which impart similar fragmentation patterns after ionization in the gas phase. This feature can be exploited by tandem mass spectrometry to specifically detect subfamilies of products. Examples of precursor ion scans, product ion scans and constant neutral loss scans will be shown that facilitate partial characterization of libraries. To solve the difficult problem of the quantitative analysis of libraries, i.e., to evaluate their equimolarity, the use of an evaporative light scattering detector (ELSD) or a chemiluminescent nitrogen detector (CLND) is suggested as more appropriate.     

Assessing the scaffold diversity of screening libraries

Medicinal chemists have traditionally realized assessments of chemical diversity and subsequent compound acquisition, although a recent study suggests that experts are usually inconsistent in reviewing large data sets. To analyze the scaffold diversity of commercially available screening collections, we have developed a general workflow aimed at (1) identifying druglike compounds, (2) clustering them by maximum common substructures (scaffolds), (3) measuring the scaffold diversity encoded by each screening collection independently of its size, and finally (4) merging all common substructures in a nonredundant scaffold library that can easily be browsed by structural and topological queries. Starting from 2.4 million compounds out of 12 commercial sources, four categories of libraries could be identified: large- and medium-sized combinatorial libraries (low scaffold diversity), diverse libraries (medium diversity, medium size), and highly diverse libraries (high diversity, low size). The chemical space covered by the scaffold library can be searched to prioritize scaffold-focused libraries.     

Combinatorial lead optimization of [1,2]-diamines based on ethambutol as potential antituberculosis preclinical candidates

Despite relatively modest potency, ethambutol (EMB, (S,S)-[N,N-di-2-amino-1-butanol]ethylenediamine) is a mainstay of contemporary chemotherapy for the treatment of tuberculosis. We have developed a solid-phase synthesis of 1,2-diamine analogues of EMB using a novel acylation-reduction sequence that is compatible with high-throughput 96-well format chemistry. Using this procedure, we have synthesized 63 238 diamine analogues in pools of 10 that are suitable for testing. MIC and a target-based reporter assay were used to direct deconvolution of 2796 individual compounds from these mixtures, and the 69 most potent molecules were resynthesized in milligram quantities for hit confirmation. Purification of these individual active diamine analogues allowed the identification of 26 compounds with activity equal to or greater than EMB. Amines which occurred most frequently in active compounds included many with large hydrophobic moieties, suggesting that optimization was perhaps selecting for the isoprenoid binding site of the arabinosyltransferase target of EMB. N-Geranyl-N'-(2-adamantyl)ethane-1,2-diamine (109), the most active of these diamines, displayed a 14-35-fold improvement in activity in vitro against Mycobacterium tuberculosis, as compared to EMB.     

Automated determination of precursor ion, product ion, and neutral loss compositions and deconvolution of composite mass spectra using ion correlation based on exact masses and relative isotopic abundances

After an accidental, deliberate, or weather-related dispersion of chemicals (dispersive event), rapid determination of elemental compositions of ions in mass spectra is essential for tentatively identifying compounds. A direct analysis in real time (DART)ion source interfaced to a JEOL AccuTOFmass spectrometer provided exact masses accurate to within 2 mDa for most ions in full scan mass spectra and relative isotopic abundances (RIAs) accurate to within 15-20% for abundant isotopic ions. To speed determination of the correct composition for precursor ions and most product ions and neutral losses, a three-part software suite was developed. Starting with text files of m/z ratios and their ion abundances from mass spectra acquired at low, moderate, and high collision energies, the ion extraction program (IEP) compiled lists for the most abundant monoisotopic ions of their exact masses and the RIAs of the +1 and +2 isotopic peaks when abundance thresholds were met; precursor ions; and higher-mass, precursor-related species. The ion correlation program (ICP) determined if a precursor ion composition could yield a product ion and corresponding neutral loss compositions for each product ion in turn. The input and output program (IOP) provided the ICP with each precursor ion:product ion pair for multiple sets of error limits and prepared correlation lists for single or multiple precursor ions. The software determined the correct precursor ion compositions for 21 individual standards and for three- and seven-component mixtures. Partial deconvolution of composite mass spectra was achieved based on exact masses and RIAs, rather than on chromatography.     

2D and 3D spatially addressed arrays for high-throughput automated synthesis of combinatorial libraries

One of the key elements in the drug discovery process is the use of automation to synthesize libraries of compounds for biological screening. The "split-and-mix" approaches in combinatorial chemistry have been recognized as extremely powerful techniques to access large numbers of compounds, while requiring only few reaction steps. However, the need for effective encoding/deconvolution strategies and demands for larger amounts of compounds have somewhat limited the use of these techniques in the pharmaceutical industry. In this paper, we describe a concept of directed sort and combine synthesis with spatially arranged arrays of macroscopic supports. Such a concept attempts to balance the number of reaction steps, the confidence in compound identity, and the quantity of synthesized compounds. Using three-dimensional arrays of frames each containing a two-dimensional array of macroscopic solid supports, we have conceptualized and developed a modular semiautomated system with a capacity of up to 100 000 compounds per batch. Modularity of this system enables flexibility either to produce large diverse combinatorial libraries or to synthesize more focused smaller libraries, both as single compounds in 12-15 micromol quantities. This method using sortable and spatially addressed arrays is exemplified by the synthesis of a 15 360 compound library.     

Use of catalyst pharmacophore models for screening of large combinatorial libraries

Using a data set comprised of literature compounds and structure-activity data for cyclin dependent kinase 2, several pharmacophore hypotheses were generated using Catalyst and evaluated using several criteria. The two best were used in retrospective searches of 10 three-dimensional databases containing over 1,000,000 proprietary compounds. The results were then analyzed for the efficiency with which the hypotheses performed in the areas of compound prioritization, library prioritization, and library design. First as a test of their compound prioritization capabilities, the pharmacophore models were used to search combinatorial libraries that were known to contain CDK active compounds to see if the pharmacophore models could selectively choose the active compounds over the inactive compounds. Second as a test of their utility in library design again the pharmacophore models were used to search the active combinatorial libraries to see if the key synthons were over represented in the hits from the pharmacophore searches. Finally as a test of their ability to prioritize combinatorial libraries, several inactive libraries were searched in addition to the active libraries in order to see if the active libraries produced significantly more hits than the inactive libraries. For this study the pharmacophore models showed potential in all three areas. For compound prioritization, one of the models selected active compounds at a rate nearly 11 times that of random compound selection though in other cases models missed the active compounds entirely. For library design, most of the key fragments were over represented in the hits from at least one of the searches though again some key fragments were missed. Finally, for library prioritization, the two active libraries both produced a significant number of hits with both pharmacophore models, whereas none of the eight inactive libraries produced a significant number of hits for both models.     

Synthesis and mass spectrometry studies of branched oxime ether libraries. Mapping the substitution motif via linker stability and fragmentation pattern.

The oxime ether chemistry has recently been used as a convenient approach to preparing potentially highly diverse combinatorial libraries. The synthetically easiest way to form the libraries is convergent, i.e., via reaction of a branched scaffold containing two or more aminooxy linker groups, with a variety of carbonyl substituents. We show here that such reactions between aldehydes and ketones of different structure with the scaffolds containing different types of aminooxy groups can lead to the formation of virtually all expected components in the model mixtures 1-3 formed from three scaffolds (7-9) and eight substituents (R(1)-R(8)). One important problem with the branched libraries is that the libraries formed from the more complex scaffolds, such as 11, contain multiple regioisomers. The results of extensive analysis of a variety of library components by mass spectrometry presented here show that the differences in the MS-MS fragmentation energies for different linkers yield regiochemical information essential for identification of individual library components.     

Monolithic silica columns for high-efficiency chromatographic separations

A deconvolution methodology for overlapped chromatographic signals is proposed. Several single-wavelength chromatograms of binary mixtures, obtained in different runs at diverse concentration ratios of the individual components, were simultaneously processed (multi-batch approach), after being arranged as two-way data. The chromatograms were modelled as linear combinations of forced peak profiles according to a polynomially modified Gaussian equation. The fitting was performed with a previously reported hybrid genetic algorithm with local search, leaving all model parameters free. The approach yielded more accurate solutions than those found when each experimental chromatogram was fitted independently to the peak model (single-batch approach). The improvement was especially significant for those chromatograms where the peaks were severely affected by the tails of the preceding compounds. Peak shifts among chromatograms, which are a usual source of non-bilinearity, were modelled in a continuous domain instead of in a discrete way, which avoided some drawbacks associated with latent variable methods. An experimental design involving simulated chromatograms was applied to check the method performance. Five main factors affecting the deconvolution were examined: concentration pattern, chromatographic resolution, number of batches and replicates, and noise level, which were evaluated using first- and second-order figures of merit. The method was also tested on three real samples containing compounds showing different overlap. Four multi-batch deconvolution methods were considered differing in the nature of the processed information and kind of peak matching among chromatograms. In all cases, the multi-batch deconvolution yielded better performance than the single-batch approach.     

On-bead combinatorial synthesis and imaging of chemical exchange saturation transfer magnetic resonance imaging agents to identify factors that influence water exchange

The sensitivity of magnetic resonance imaging (MRI) contrast agents is highly dependent on the rate of water exchange between the inner sphere of a paramagnetic ion and bulk water. Normally, identifying a paramagnetic complex that has optimal water exchange kinetics is done by synthesizing and testing one compound at a time. We report here a rapid, economical on-bead combinatorial synthesis of a library of imaging agents. Eighty different 1,4,7,10-tetraazacyclododecan-1,4,7,10-tetraacetic acid (DOTA)-tetraamide peptoid derivatives were prepared on beads using a variety of charged, uncharged but polar, hydrophobic, and variably sized primary amines. A single chemical exchange saturation transfer image of the on-bead library easily distinguished those compounds having the most favorable water exchange kinetics. This combinatorial approach will allow rapid screening of libraries of imaging agents to identify the chemical characteristics of a ligand that yield the most sensitive imaging agents. This technique could be automated and readily adapted to other types of MRI or magnetic resonance/positron emission tomography agents as well.