High throughput HPLC/MS purification in support of drug discovery

Abstract

High throughput purification techniques are an important part of drug discovery and provide high-quality compounds for biological screening. In this paper, we describe the purification platform developed by ArQule that is based on reverse phase high performance liquid chromatography (RP-HPLC) separation and mass directed fractionation. By strictly enforcing collection of only one fraction per sample, this purification paradigm has significantly enhanced the throughput and simplified the post-purification operation. Recovery studies have proven the reliability of this process and development of fast chromatographic separations provide enhanced throughput without additional capital investment. This approach has been used successfully to purify over half a million compounds in the past 2 years and resulted in post-purification average purity of over 97% when assessed by HPLC and low-wavelength UV. © 2004 Elsevier B.V. All rights reserved.

Introduction In the early 1990s, split-and-combine combinatorial chemistry emerged as an innovative synthesis approach and promised to revolutionize drug discovery by delivering mixtures of large numbers of small molecules [1,2]. The expectation was that the large number of analogs synthesized and screened against emerging novel targets provided by the genomic revolution would interrogate the biologically active chemical space much more effectively, and thus, quickly lead to new drug candidates. In general, however, the rapid generation of mixtures was not able to deliver on this promise. While many new chemical entities were synthesized, the fact that they were exposed to the targets as mixtures has led to difficulties in de-convoluting active structures and a significant need for resources to follow up on many false actives. High throughput parallel synthesis was developed around the same time and promised to improve the productivity of drug discovery based on a different approach [3]. In parallel synthesis, individual molecules are synthesized in a spatially addressable format and these compounds can be screened similar to individual compounds made by medicinal chemists. Relying heavily on automation for leverage, this approach can generate many more compounds compared to the traditional “one at a time” approach practiced by medicinal chemists.

Although only one molecule is intended to be synthesized per reaction vessel, almost inevitably, there are impurities present at the end of the reaction, especially when multiple synthetic steps are employed to generate more complex drugor natural product-like structures. To assess the biological results and develop meaningful structural activity relationships (SAR) that can be used to guide lead optimization, compound characterization relative to purity and quantity is imperative. There have been many different approaches in assessing the purity of parallel synthesis products. While nuclear magnetic resonance (NMR) remains the “gold standard” for traditional medicinal chemists, currently the cost and complexity of automated NMR data analysis and interpretation prevent its application as a high throughput process. Library compounds made by parallel synthesis are predominantly analyzed using high performance liquid chromatography (HPLC) with UV and/or evaporative light scattering detectors (ESLD) for purity with on-line mass spectrometry (MS) detection for structure confirmation [4–7]. This approach, however, does have limitations since none of these detectors provide a truly “universal” response for all reagents, intermediates and final products. The inherent bias of each detector such as extinction coefficients for UV, vapor pressure for ELSD or ionization efficiency for MS for each sample constituent renders a very complex analytical problem.