Dunayevskiy and Hughes [84] proposed the ultimate in multidimensional chromatographic methods for the screening of drug candidates noncovalently bound to receptors by combining serially GPC, ultrafiltration, and RP-HPLC

in on-line and off-line modes with ESI-MS detection in a presumably high throughput format. The on-line mode incorporates column switching and sample trapping, while the off-line mode incorporates a lyophilization-concentration-resuspension step. An alternative unreported approach would be to replace the ultrafiltration step with turbulent flow chromatography to remove the high mass receptor (protein) from the low mass ligands.

Capillary Electrophoretic/ESI-MS Methods: Affinity Capillary Electrophoresis/ESI-MS and Capillary Isoelectric Focusing/ESI-MS

Capillary electrophoresis (CE) separations are produced by differential migration of solutes confined to narrow-bore capillaries subjected to high electric fields. The mobility of a species is related directly to its net charge and is inversely related to its hydrodynamic drag (mass and shape). Affinity capillary electrophoresis (ACE) [85] is achieved when a migration change for a ligand (or receptor) occurs due to the interaction with a receptor (or ligand) present in the electrophoretic buffer. A procedure often used in ACE studies is to maintain in the electrophoretic buffer the receptor as a neutral species (by adjusting the pH to the isoelectric point (pI) of the receptor) and the ligands as charged species. The mobility of the ligand-receptor complex is lower than the free ligand due to the additional hydrodynamic drag of the complex vs. the free ligand, while the net charges of both species are identical. This approach is very useful for screening components of chemical libraries, which selectively bind to receptors due to the high resolution and high sensitivity achievable with ACE. The most common in-line detectors used have been UV-Vis and laser-induced fluorescence; however, when the column output is interfaced to an ESI-MS, the resolved components could be identified by their molecular weights and their structures elucidated by MS/MS.

Karger and coworkers [86-88] demonstrated the use of ACE/ESI-MS for the identification of new peptide constructs that bind more strongly to van-comycin than the known ligand D-Ala-D-Ala (AA). The procedure used was to adjust the electrophoretic buffer pH to that of the pI of the vancomycin receptor (pI 8.1). The ligands used were members of a peptide library with the structure Fmoc-DDX1X2, where X was one of 10 selected D-amino acids. Figure 2.10A and 2.10B, respectively, illustrate electropherograms obtained without and with vancomycin in the electrophoretic buffer. Based on the ESI-MS analysis, Figure 2.10A exhibits three distinct peaks (I, II, III), which from ESI-MS analysis contained peptides with two, one, and no glutamic acid residues corresponding to net charges of -5, -4, and -3, respectively. Figure 2.10B exhibits a similar group of three peaks, but shows in addition a series of peaks with longer migration times corresponding to noncovalent

Migration Chromatographie
FIGURE 2.10 ACE-UV electropherograms of the Fmoc-DDXi X2 library (100 peptides) (A) without, and (B) with vancomycin. See text for discussion. (Reprinted from Chu et al. [87], used with permission. Copyright 1996 by the American Chemical Society.)

vancomycin-Fmoc-DDX1X2 complexes, where the residues X1 and X2, listed in Figure 2.10B, were identified by ESI-MS and ESI-MS/MS. The interesting feature is that a number of peptides have migration times greater than that of AA corresponding to binding affinities greater than that of AA. In addition, these stronger binding peptides, all C-terminating with A, have penultimate C-terminal residues F, Y, and H, each containing aromatic moieties. ACE/ESI-MS has also been used for evaluating the binding to vancomycin of larger peptide libraries, and it appears that the upper limit in a single assay of this type is about 1000 peptides. To extend such ACE/ESI-MS assays to even larger numbers of components, prescreening of the mixtures prior to ACE with affinity columns [87] or affinity coated magnetic beads [89] has been demonstrated. This approach for in vitro exploratory and early-discovery drug screening appears to be very promising due to the small quantities of precious receptors needed, the high resolution achieved, and the short analysis times required. However, the sensitivities achieved with non-mass-spectrometric detectors and their reliability and robustness is greater than that often achievable with ESI-MS. In general, special ACE methods have to be developed for the analysis of receptor complexes formed with uncharged ligands. Other ACE/ESI-MS applications, potentially useful for drug screening, are epitope mapping [90], analysis of DNA complexes [91], FK506-Binding Protein [FKBP]-rapamycin complexes [92], and binding constants of vancomycin-peptide complexes [93].

Recently, capillary isoelectric focusing/ESI-MS has been demonstrated as a potential method for in vitro screening of compounds that noncovalently bind to active proteins [94,95]. The procedure used for this assay is to load into a capillary a mixture of the receptor of interest (generally a protein) ligands, and an ampholyte. Upon applying a high voltage across the capillary, a pH gradient is formed and a current is generated which decreases as the components separate and focus at their isoelectric points. This focusing step can concentrate the sample up to two orders of magnitude. At this point, the resolved components, which could include the receptors, ligands, and receptor-ligand complexes, can be mobilized by applying a low external static pressure to transport the resolved components from the capillary to an ESI-MS for detection.

Tethering with HPLC/ESI-MS Detection

An ingenious screening method was developed using the covalent disulfide exchange reaction between members of a disulfide-containing chemical library with a native or engineered cysteine in the region of the active site of an enzyme protein [96,97] or the surface region of a protein-protein interaction [98,99]. The formation of the disulfide bond with the protein, referred to as tethering, is stabilized by the noncovalent interaction of the moiety attached to the disulfide with the active site in the protein. To evaluate the stability of the protein-disulfide product, the reaction was conducted under reducing conditions (1 mM 2-mercaptoethanol) in buffer under native conditions. The protein-captured disulfide ligands were identified with a ballistic gradient HPLC/ESI-MS method [100,101]. Libraries of disulfide compounds were screened in this manner as mixtures of 10 compounds. Approximately 2.5 million compounds were analyzed annually with this methodology to screen for small-molecule antagonists of enzyme and protein-protein interactions. The tethering methodology has also been extended to search for more than one adjacent active site in a protein with HPLC/ESI-MS for the detection of the covalently bound disulfide protein-ligand product [102].

Condensed-Phase Competitive Binding Assays

Two approaches using mass spectrometry have been reported to determine the binding affinities of ligands with proteins in the condensed phase. One method determines the relative binding affinities by analyzing the GPC-spin-column eluent of a mixture of ligands incubated with the protein by determining the abundances of the individual components [49]. The ESI-MS response of each of the ligands present in the eluent, normalized to the response of a fixed concentration of each of the ligands, should be proportional to the relative binding affinities for each of the ligands. A more sophisticated approach was taken by Wanner and co-workers [103-105]. A compound known to have a high binding affinity with a protein was incubated with a ligand of unknown binding affinity. Various concentrations of the ligand of unknown binding affinity were incubated with the strongly bound ligand-protein complex. For each sample, the protein was removed by ultracentrifugation of filtration and from the measured mass-spectral response for the strong binder, the concentration of the competing drug that inhibits 50% of the specific binding (IC50) is computed. This approach mimics radiological assays [106] commonly used for determining IC50, but is simple and straightforward without requiring radioligands.


MALDI-MS has traditionally been used extensively for the analysis of pep-tides, proteins, and oligonucleotides. To aid in these assays, a number of researchers, including Hutchins [107-112], Nelson [113,114], Kris [115-117, 118], Weinberger [119], and their co-workers, have developed MALDI probe tips that have been modified by activating the metal surface with a chemical cross-linker that was covalently linked with active biomolecules. These active biomolecules serve as affinity targets for solutions containing ligands to which the surface is exposed. These ligands can then be released and characterized directly by MALDI-MS upon UV-laser irradiation in the presence, or even the absence, of an UV-absorbing matrix. In addition, the probe surface can be subdivided into smaller subregions, each individually activated with a biomolecular target, for high throughput analysis [120]. An extension of this approach is the use of the affinity surface of a surface-plasmon resonance spectroscopy Biacore chip as the probe surface for MALDI-MS [121-123]. Even though the use of affinity MALDI probes and Biacore chips with MALDI-MS detection show great promise as tools for high throughput in vitro bioaffin-ity drug screening for small molecules, no examples have appeared in the literature at this time.

An example of chemical screening with a modified surface of a MALDI-MS probe was described recently [124]. The gold surfaces of the MALDI plate wells were prepared with a self-assembled monolayer (SAM), formed from a peptide-terminated alkane thiolate. The free peptide terminus, a substrate for anthrax lethal factor, is cleaved by the anthrax enzyme in the absence of a chemical inhibitor. Cocktails consisting of mixtures of eight compounds with anthrax lethal factor were applied to each MALDI plate well containing the SAM. After incubation, the wells were washed, and MALDI mass spectra were obtained. Inhibitors of anthrax lethal factor exhibited spectra in which the peptide-terminated alkane thiolate was intact after the enzymatic reaction. Using this MALDI-MS technique, the rapid screening of chemical libraries for inhibitors of anthrax lethal factor was demonstrated.

Fluorescence-Activated Cell Sorting (Flow Cytometry)/MALDI-MS

Keough and colleagues [125,126] developed a sophisticated method for screening support-bound combinatorial libraries of pharmaceutical interest by using affinity methods coupled with flow cytometry for selecting beads with active components and MALDI-MS for structural identification. By taking advantage of the mass-spectral properties of MALDI samples, support-bound combinatorial libraries were prepared, which, upon MALDI analysis, revealed the chemical structures of the active components. A series of moieties is sequentially linked on a bead by the "randomize and split" method, such that a combinatorial library is generated where each individual bead has one unique structure. The library is designed such that the initial moieties are chemically or photochemically sensitive so that the compound can be easily cleaved from the bead for MALDI analysis. In addition, during each step in the combinatorial synthesis, a small fraction of the growing sequence is capped to produce a mixture of related terminated components on the bead. Capping agents can be used to mass code each step in the synthesis to indicate, for example, the use of D- or L-amino acid residues or to differentiate between structural isomers. Upon cleaving the components from a single bead, the sequence of the compound can be read directly from the mass differences in the molecular ions of the capped components generated in the MALDI mass spectrum. The structure could then be determined unambiguously from the unique masses of the moieties used in the synthesis.

Support-bound combinatorial libraries can be prepared with the chemical properties, as described earlier. Affinity methods were developed where soluble fluorescently labeled proteins were mixed with the library and nonco-valently bound to the minority of beads containing active sequences. Using a flow cytometer, the beads with fluorescently bound protein can be sorted from the majority of beads that do not contain the active sequence and the compounds on the active beads sequenced by MALDI-MS. In a model study, an original million-member peptide library was screened against an anti-HIV-1 gp 120 monoclonal antibody. Seven beads with the highest fluorescence were found to have the predicted consensus sequence, as measured by MALDI-MS.

Similar studies with MALDI-MS to sequence members of support-bound combinatorial libraries of peptides [127,128], glycopeptides [129], phos-phinic peptides [130], cyclic oligocarbamates [131], and antisense DNA oligonucleotides [132], all manually selected after incubation with fluores-cently labeled proteins, have been reported. Such studies were also proposed for screening support-bound combinatorial libraries of small molecules with a central scaffold for pharmaceutical applications [125]. In all these screening procedures, fluorescence and MALDI methods are coupled together such that fluorescence methods are used to identify the active components and MALDI-MS essentially plays the role of elucidating the structures of the active components. ESI-MS could also be used to analyze the drugs once cleaved from the beads; however, this may require additional sample manipulation steps to those needed for MALDI-MS.

MALDI-MS Analysis of Biopolymer-Ligand Complexes

Direct analysis of noncovalent biopolymer-ligand complexes by MALDI-MS is difficult [133,134]. The complexes must be prepared in the condensed phase, in a buffer system that maintains the native state, which remains intact upon crystallization and in the gas phase during MALDI analysis. Many factors are involved in preparing the samples, including choice of matrix, matrix-to-analyte concentration, receptor and ligand concentrations, pH, buffers, solvents, and ionic strength. Best results have been obtained with neutral matrixes such as 6-aza-2-thiothymine, 2,6-dihydroxyacetophenone, and neutralized sinapinic acid for UV-MALDI. Good-quality MALDI spectra of noncovalent complexes often are obtained only from the summed first laser shot spectra acquired from regions of the sample not previously irradiated. A number of noncovalent protein-ligand complexes have been studied by UV-MALDI-MS, including RNase S (noncovalent complex of S-protein and S-peptide) [135], myoglobin-heme complex [136], adenylate kinase complexes with AMP, ADP, and ATP [137], Ras-GDP, and Ras-GppNp [guanosine-5'-(^, Y-imido)-triphosphate] [138]. Recently, the noncovalent complex of vancomycin with Ac2-L-Lys-D-Ala-D-Ala was observed in the positive-ion mode by IR-MALDI, but not by UV-MALDI [139]. However, in a control study, where vancomycin was prepared with a 1:1 mixture of Ac2-L-Lys-D-Ala-D-Ala:Ac2(d6)-L-Lys-L-Ala-L-Ala, the negative ion IR-MALDI spectrum surprisingly exhibited the vancomycin complexes with an intensity ratio of 1:3, suggesting a considerable difference between the condensed phase and MALDI gas-phase binding affinities, since no complex with the

Ac2(d6)-L-Lys-L-Ala-L-Ala isomer was expected. At this point, no high throughput direct-screening studies of biopolymers noncovalently bound to drug candidates with UV- or IR-MALDI-MS have been reported. A good starting point for drug screening with UV- and IR-MALDI-MS would be screening for cell-wall inhibitors, as described earlier for the direct analysis of vancomycin-peptide complexes with ESI-MS. In addition, it would be of great interest to compare the MALDI-MS results, where the desorption and ionization steps occur nearly simultaneously, with those obtained with the two-laser mass spectrometry methodology (L2-MS) [140], where the neutral complex is sequentially desorbed and ionized with IR and UV lasers, respectively, prior to mass spectral analysis.

The MALDI-MS method can also be used to analyze the small molecules isolated in the condensed-phase separation techniques just described for resolving biopolymer-drug complexes from unreacted drug or biopolymer. The chromatographic methods, with ESI-MS as a flow-analysis detector, could as well have used MALDI-MS as an off-line detector. However, with the exception of peptides and nucleotides, MALDI-MS has not become a popular detector for screening small molecules, because of the high matrix and chemical background at low mass, the possibility for fragmentation of the small molecule, and the possible reaction of the analyte with the MALDI matrix.

Hydrogen/Deuterium Exchange for Screening Protein-Ligand Complexes with MALDI-MS Detection

Fitzgerald and co-workers [141-145] developed a method for screening the stability of a protein-ligand complex by determining the extent of hydrogen/ deuterium (H/D) exchange for the protein in the presence of a ligand as a function of titrated denaturing agent. The mass change of the protein, monitored by MALDI-MS, demonstrated that the lower the extent of H/D exchange, the stronger the ligand bound, corresponding to a lower protein-ligand dissociation constant. This methodology was extended to a single-point measurement, which was proposed as a promising method for high throughput screening for identifying protein-ligand interactions [146].

Was this article helpful?

0 0

Post a comment