Info

tions of mass transfer, all displacements or breakthroughs would be rectangular in shape. Under real-world conditions where ligand-binding kinetics dominate, the resulting finite mass transfer rates serve to "smooth" these features into peaks and sigmoids [8].1

FAC has a remarkable capacity to resolve/rank-order ligands even under fully saturating levels of compound. The example in Fig. 6.7 shows the evolution of multiple-ligand breakthroughs under conditions where the cartridge saturation is over 99.9%. A mixture of eight ligands generates a rank order as determined by measurement of breakthrough volumes, which follows the general trend of IC50 values (3.3 mM to 39 nM; Table 6.1).

Note that the mixture contains the three isobaric ligands discussed above, and even under these saturating conditions, the method can resolve three orders of magnitude Kd values. A multi-ligand, equilibrated environment supports the Kd measurement of both the strongest and the weakest ligand, which provides the opportunity to "bracket" the rank order with accurate dissociation constants [19]. In theory, the ligands with intermediate binding strength could also be quantitated, requiring knowledge of ligand concentration and a measurement of rollup amplitude [11]. This may be possible in certain applications (e.g. screening of mixtures constructed from single compound stock solutions) but, in more challenging screenings as encountered using natural product extracts or crude mixture syntheses, a bracketed rank order is the best that can be achieved.

Deng and Sanyal have suggested that FAC is not applicable to ligands with low on- and off-rates [20]. As will be shown below, this is not true particularly when indirect FAC methods are applied, but it is also misleading in direct assays as described in this section. With flow-rate programming, for example, slow kinetics

1) These displacements (or ''roll-ups'') can only occur if there is competition for a common binding site, or a strongly-negative allosteric linkage between two distal sites, with the further requirement that the allosteric effector have a lower Kd.

Time (min)

Fig. 6.8 Breakthrough curves for ligands infused through an estrogen receptor b FAC assay. Slow tight-binding ligands (nafoxidine and tamoxifen) exhibit diffuse breakthrough curves, while ligands with rapid kinetics exhibit sharper curves (norethindrone). Dehydroisoandro-sterone, a ligand intermediate between norethindrone and nafoxidine, was undetected in this experiment.

Time (min)

Fig. 6.8 Breakthrough curves for ligands infused through an estrogen receptor b FAC assay. Slow tight-binding ligands (nafoxidine and tamoxifen) exhibit diffuse breakthrough curves, while ligands with rapid kinetics exhibit sharper curves (norethindrone). Dehydroisoandro-sterone, a ligand intermediate between norethindrone and nafoxidine, was undetected in this experiment.

need not be limiting. Nevertheless it is true that any flowing system must consider the impact of mass transfer rates on the quality of the data. As an example, consider Fig. 6.8. Here, we screened a mixture of @100 compounds in an estrogen receptor b FAC assay. The two stronger ligands (nafoxidine and tamoxifen) are known to possess extremely slow off-rates yet the FAC method not only detects these as ligands, it also correctly estimates their ranking. Under the temporal conditions of this experiment, neither nafoxidine nor tamoxifen rapidly establish equilibrium on the FAC cartridge. In this situation, ligands with slower on-rates will elute ahead of those with faster on-rates. Notice also the severely extended breakthrough curves for both ligands, indicative of slow on-rates [9]. So rather than an inherent limitation, the FAC-MS method is useful for detecting slow kinetics in much the same way as optical biosensors. Admittedly, theoretical treatments of breakthrough curves for on- and off-rate measurements have not yet received wide application to biomolecular interaction data from FAC. It is also true that, if on-rates are very low (<100 M-1 s-1), the direct FAC method may miss them. But for the simple purpose of ligand discovery, we suggest that FAC cartridges be operated at low linear flow rates without concern that slow/ tight-binding ligands may be missed (this will be revisited in the next section). Notice that the FAC data in this example only shows breakthrough curves for three of the four expected ligands. Dehydroisoandrosterone is undetectable at the concentrations chosen, but the weakest of the four ligands (norethindrone) has been displaced - clearly indicating the presence of another ligand and indicating the utility of the displacement as a useful check for completeness of the analysis.

Mass spectrometry enables the type of direct analyses described, but it does have its limitations. Online operation forces detection at infusion concentrations, in salty buffer and under complex mixture conditions. General ion suppression results from the buffer and mixture components, and mixture complexity can tax the resolution of even the best mass spectrometers. Increasing compound concentration is not the answer, as this leads to problems of solubility and increased compound consumption. We have found that the online method can work successfully for up to 100 compounds per analysis, but the false negative rate becomes appreciable [21]. As an alternative for ligand discovery purposes, we have developed a FAC-LC/MS system in which FAC effluent is sampled and analyzed by LC/MS [19]. This system offers the ability to concentrate mixture components and introduces another dimension to the data in order to tolerate more complex mixtures (Fig. 6.9). Using this system, we have screened approximately 1000 modified trisaccharide acceptor analogs targeting immobilized N-

Fig. 6.9 Schematic of FAC effluent sampling strategy for insertion of an LC/MS step to increase ruggedness of the discovery mode of analysis, as applied to high throughput screening for ligands to GnT-V [19]. The insets represent LC/MS data for a strong ligand for four fractions of FAC effluent (1, x, x + 1, x + 2). Blue traces represent the extracted ion chromatogram of the compound eluted from the GnT-V column, and red indicates the extracted ion chromatogram (XIC) from a blank column. The insets are referenced to idealized online FAC-MS chromatograms for the strong ligand. Adapted with permission from the American Chemical Society.

Fig. 6.9 Schematic of FAC effluent sampling strategy for insertion of an LC/MS step to increase ruggedness of the discovery mode of analysis, as applied to high throughput screening for ligands to GnT-V [19]. The insets represent LC/MS data for a strong ligand for four fractions of FAC effluent (1, x, x + 1, x + 2). Blue traces represent the extracted ion chromatogram of the compound eluted from the GnT-V column, and red indicates the extracted ion chromatogram (XIC) from a blank column. The insets are referenced to idealized online FAC-MS chromatograms for the strong ligand. Adapted with permission from the American Chemical Society.

Table 6.2 Summary of the hit data obtained from the GnT-V screening using the FAC-LC/MS method [19]. Hits are categorized based on an arbitrary binning strategy, where breakthrough timesa are converted into approximate Kd values. Adapted with permission from the American Chemical Society.

Rank order Number of Hits in Library Approximate

(1 - weak, to 4 - strong) (~1000 compounds) Kd Values (mM)

acetylglucosaminyltransferase V (GnT-V), an enzyme regulating the branching pattern of N-linked oligosaccharides on glycoproteins [22]. Increased expression of active GnT-V has been reported in mammary, hepatocellular, and pancreatic cancer [23, 24] and it has been suggested that inhibition of GnT-V could represent a useful treatment for cancer [25]. With an LC/MS system incorporating hydro-philic interaction chromatography (HILIC) for oligosaccharide separation, we have discovered four quality ligands (with Kds in the 0.6-1.5 mM range, measured in separate FAC-MS experiments), from a large mixture possessing an overall hit rate of approximately 5% (Table 6.2).

Fractionating the FAC effluent for LC/MS processing reduces resolution in hit determination, as the breakthrough volumes for each hit can only be estimated, however coarse fractionation will suffice for screening purposes. The insertion of an LC step avoids a dependency on MS-compatible assay buffers and provides an increase in sensitivity along with the ability to detect lower concentrations of ligand; this is important when screening large mixtures as high total library concentrations can induce compound precipitation. Based on this work, screening rates exceeding 5000 compounds day-1 are easily achieved with a simple LC/MS system, at compound concentrations 10- to 100-fold less than the online method. With automation and more highly resolving LC/MS systems, these screening rates could easily exceed 50 000 compounds day-1, assuming mixtures of approximately 5000 compounds each.

There are really only two situations in modern drug discovery where this sort of capacity for high-volume mixture screening may be needed. First, natural product extracts can be screened to detect low-abundance compounds present in the complex matrix of nonligand species. Some interesting work has been published in this area [26], and will be described below. Second, split-pool synthetic combinatorial libraries can be quickly surveyed, possibly as a method for surveying chemical space prior to launching an expanded parallel-chemistry effort for library creation. Blending pre-existing collections of individual compounds is possible in order to take advantage of the efficiency of the screening method, however these benefits may be eroded due to the effort in reformatting compound collections at time of use. Given the pre-existing investment in large-scale screening systems, the niche for FAC-MS in a screening laboratory is likely before and after the main screening exercise - during library development, and hit evaluation or secondary screening.

Was this article helpful?

0 0

Post a comment