Analysis and screening of combinatorial libraries using mass spectrometry

Transcription

1 BIOPHARMACEUTICS & DRUG DISPOSITION Biopharm. Drug Dispos. 22: (2001) DOI: /bdd.278 Analysis and screening of combinatorial libraries using mass spectrometry Young Geun Shin and Richard B. van Breemen* Department of Medicinal Chemistry and Pharmacognosy, College of Pharmacy, University of Illinois at Chicago, 833 S. Wood Street, Chicago, IL 60612, USA ABSTRACT: Mass spectrometry is a highly selective and high throughput analytical technique that is ideally suited for the identification and purity determination of large numbers of compounds prepared using combinatorial chemistry or for the dereplication of natural products. Compounds may be characterized based on molecular weight, elemental composition and structural features based on fragmentation patterns. When coupled to a separation technique such as highperformance liquid chromatography (HPLC) or capillary electrophoresis, mass spectrometric applications may be expanded to include analysis of complex mixtures. However, the slower speed of the separation step reduces the throughput of the analysis. This review concerns the application of mass spectrometry to the characterization of combinatorial libraries and the screening of library and natural product mixtures. Strategies to enhance the throughput of LC MS are discussed including fast HPLC and parallel LC MS. Also, mass spectrometry-based screening methods are described including frontal affinity chromatography mass spectrometry, gel permeation chromatography LC MS, direct electrospray mass spectrometry of receptor ligand complexes, affinity chromatography mass spectrometry, and pulsed ultrafiltration mass spectrometry. Copyright # 2001 John Wiley & Sons, Ltd. Key words: mass spectrometry; high throughput screening; combinatorial chemistry; drug discovery; LC MS Introduction One of the most active areas of pharmaceutical research is focused on accelerating the pace of * Correspondence to: Department of Medicinal Chemistry and Pharmacognosy, College of Pharmacy, University of Illinois at Chicago, 833 S. Wood Street, Chicago, IL 60612, USA. Breemen@uic.edu Contract/grant sponsor: National Cancer Institute; contract/grant number: R24 CA83124 Contract/grant sponsor: Office of Dietary Supplements (ODS); contract/grant number: p50 AT00155 Contract/grant sponsor: National Institute of General Medical Sciences (NIGMS) Contract/grant sponsor: Office for Research on Women s Health (ORWH) Contract/grant sponsor: National Center for Complementary and Alternative Medicine (NCCAM) drug discovery. By more rapidly identifying, optimizing and developing new chemical entities from among diverse libraries of synthetic compounds or natural products, new drugs may be brought to market faster than ever before. Many approaches have been applied and developed to address this goal, and during the last 10 years, two new paradigms have emerged that are making a significant impact; combinatorial chemistry and high throughput screening (HTS). The application of combinatorial chemistry [1, 2] and high throughput screening [3] to drug discovery has altered the traditional serial process of lead identification and optimization that traditionally preceded investigations of drug development. Instead, vast numbers of new compounds are synthesized in parallel using Copyright # 2001 John Wiley & Sons, Ltd.

2 354 Y.G. SHIN AND R.B. VAN BREEMEN combinatorial chemistry and tested rapidly using high throughput screening. As a result, the synthesis of new chemical entities is no longer the rate-limiting step during drug discovery. Instead, a new bottleneck has become the analytical verification of the structure and purity of each library compound synthesized using combinatorial chemistry or of each lead compound obtained using high throughput screening. Consequently, more rapid and robust analytical approaches are needed to keep pace with these new paradigms. Traditionally, spectroscopic and chromatographic techniques such as infrared spectroscopy, nuclear magnetic resonance spectroscopy, fluorescence spectroscopy, gas chromatography, highperformance liquid chromatography (HPLC) and mass spectrometry have been used to support drug discovery and development. Although these techniques still are being used to various extents to support combinatorial chemistry and high throughput screening, some of them such as gas chromatography and fluorescence spectroscopy are not applicable to most new chemical entities, some are not specific enough for chemical identification (e.g. infrared spectroscopy) and other techniques suffer from low throughput (e.g. nuclear magnetic resonance spectroscopy. Unlike gas chromatography, HPLC is compatible with almost all drugs or a drug-like molecule without the need to derivatize the compounds prior to analysis. In addition, mass spectrometry provides a universal means to characterize and distinguish drugs based on both molecular weight and structural features while at the same time providing high throughput. With the development of routine LC MS interfaces and ionization techniques such as electrospray and atmospheric pressure chemical ionization (APCI), mass spectrometry has also become an ideal HPLC detector [4]. Electrospray and APCI mass spectrometry, LC MS, tandem mass spectrometry (MS MS), and LC MS MS have become fundamental tools in the analysis of combinatorial libraries and subsequent drug development studies [5 7]. Since ionization techniques such as electrospray and APCI tend to produce abundant molecular ions but little or no fragment ions, MS MS with collision-induced dissociation (CID) is often used to promote fragmentation of molecular ion precursors and then record their product ions. Such tandem mass spectra contain structurally significant fragment ions that provide information regarding the identity and in many cases uniquely identify the sample under study. In this review, applications of mass spectrometry will be discussed for the analysis of combinatorial libraries, the screening of combinatorial libraries and natural product extracts, and the identification of pharmacologically active leads. Mass spectrometry-based HPLC purification of combinatorial libraries Although combinatorial libraries may be synthesized as mixtures, most libraries are prepared in parallel as discrete compounds on the milligram scale. In order to assure the validity of subsequent screening analysis, many researchers prefer to verify the structure and purity of each compound prior to high throughput screening. Several strategies have been used to purify milligram quantities of compounds including crystallization, preparative thin-layer chromatography, and semi-preparative HPLC. Among these, semi-preparative HPLC is gaining popularity for the purification of combinatorial libraries because of high throughput and the ease of automation. Typically during semi-preparative HPLC, fraction collection is initiated whenever a UV signal is observed above a predetermined threshold. This procedure usually results in the collection of several fractions per analysis and hence creates additional issues. For example, the need to collect multiple fractions per sample means large fraction collector beds or assays will be required when purifying large numbers of compounds. Furthermore, secondary analysis of each fraction must be carried out using flow-injection mass spectrometry, LC MS, or LC MS MS to identify the appropriate fractions. When purification of large numbers of combinatorial libraries is required, this approach can become prohibitively time consuming. In order to carry out the purification of compound libraries more efficiently and rapidly, the steps of HPLC purification and mass spectrometric analysis may be combined into automated mass-directed fractionation [8 10]. Depending on

3 MASS SPECTROMETRY FOR HIGH THROUGHPUT SCREENING 355 the scale required, various HPLC column sizes may be selected such as analytical, semi-preparative, or preparative separation. All components including autosampler, injector, HPLC, switching valve, mass spectrometer and fraction collector are controlled by computer, so that the procedure is fully automated. In order to minimize the number of fractions, the system may be programmed to collect only those peaks displaying the desired molecular ions, protonated or deprotonated molecules, etc. Alternatively, all peaks displaying abundant ions within a specified mass range may be fraction collected. An example of the mass-directed purification of a compound prepared during the parallel synthesis of a combinatorial library of discrete compounds is shown in Figure 1. In this case, the crude reaction product is only 30% of the mixture (Figure 1A). However, the expected product is easily detected by its molecular ion in the computer-reconstructed ion chromatogram (Figure 1B). After mass-guided fractionation, reanalysis using LC MS showed that the desired product was >90% pure (Figure 1C). The use of mass spectrometry-based purification of combinatorial libraries provides a means for reducing the number of HPLC fractions collected per sample. Since mass spectra may be stored corresponding to each fraction, this approach also eliminates the need for postpurification analysis to further characterize and identify each compound as would be necessary when using UV-based fractionation. However, this approach still requires the expenditure of time for HPLC and subsequent removal of solvent from each fraction prior to storage or screening. Although the selection of the HPLC mobile phase is limited to mass spectrometry compatible solvents, this is usually not a concern. However, care should be used to insure that the mobile phase and ionization mode are compatible with the desired product. Otherwise, molecular ion species might not be formed. Alternatively, supercritical fluid chromatography mass spectrometry (SFC MS) has been used for the high-throughput analysis of combinatorial libraries [11, 12], and it offers benefits for use during mass spectrometry-based purification of combinatorial libraries. The advantages of SFC MS relative to conventional LC MS are (1) lower viscosities and higher diffusivities of condensed CO 2 than HPLC mobile phases, and Figure 1. Mass-directed purification of a combinatorial library. Chromatographic separation was carried out using gradient elution of 10 90% acetonitrile in water for 7 min following an initial hold at 10% acetonitrile for 1 min. (A) Total ion chromatogram showing desired product and impurities. (B) Computer-reconstructed ion chromatogram (RIC) corresponding to the expected product. (C) Post-purification analysis of the isolated component with a purity >90%. (Reprinted from Reference [8], J Chromatogr A 1998; 794 (1 2): 3 13 with permission from Elsevier Science.)

4 356 Y.G. SHIN AND R.B. VAN BREEMEN (2) ease of solvent removal and disposal after analysis. However, SFC instrumentation remains more expensive and less widely available than conventional HPLC systems. Mass spectrometric confirmation of molecular weight, elemental composition, structure and purity of combinatorial compounds The confirmation of molecular weight and elemental compositions of compounds used for high throughput screening, whether discrete compounds or combinatorial library mixtures, is typically carried out using mass spectrometry. The throughput of classical gravimetric methods (e.g. combustion analysis) for the determination of elemental compositions is too slow to keep pace with combinatorial chemical synthesis. Also, gravimetric methods require pure compounds and are incompatible with mixtures. In addition to the measurement of exact mass for composition determination, mass spectrometry may be used to assess the purity of compounds prepared by organic synthesis or isolated from natural sources. Molecular weight measurements may be carried out using virtually any mass spectrometer equipped with suitable ionization techniques. For the analysis of most drugs and drug candidates including products of combinatorial chemistry, electrospray ionization and APCI are used. The simplest and highest throughput use of mass spectrometry for confirming molecular weights and purity of drug candidates is flow injection analysis of sample solutions using electrospray or APCI mass spectrometry. Unlike classical electron impact or chemical ionization, APCI and electrospray typically form abundant molecular ions, protonated molecules, or deprotonated molecules of drug-like compounds without prior derivatization. Since no sample preparation is necessary, the throughput of the analyses is enhanced over classical mass spectrometric methods. Since the presence of multiple compounds or even buffers can interfere with ionization of some compounds, HPLC separation is sometimes required prior to mass spectrometric analysis. Fortunately, electrospray and APCI are also popular interfaces between HPLC systems and mass spectrometers so that LC MS may be carried out as a single analysis instead of in separate steps. In addition to the determination of molecular weight which is the standard function of organic mass spectrometers, high-precision mass spectrometers may be used for the measurement of exact masses of drugs and drug candidates for the determination of elemental compositions. Expanding performance even further, high resolution in combination with high precision is especially useful for determining the elemental compositions of compounds in combinatorial library mixtures without having to isolate each compound using chromatography or some other separation technique. High resolution is the ability of the mass spectrometer to separate ions differing by only a small fraction of a mass unit. In mass spectrometry, resolution is defined as M=DM, where M is the m=z value of a singly charged ion and DM is the difference (measured in m/z) between M and the next highest ion. Alternatively, DM may be defined in terms of the width of the peak. Examples of high resolution mass spectrometers used for exact mass measurements include double focusing magnetic sector mass spectrometers, reflectron time-of-flight (TOF) mass spectrometers, hybrid quadrupole TOF instruments, and Fourier transform ion cyclotron resonance (FT ICR) mass spectrometers. Some recent applications of FT ICR mass spectrometers to combinatorial library analysis are discussed in the next section. Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry FT ICR mass spectrometry is unsurpassed in terms of resolution (often exceeding ) and mass accuracy. In contrast to quadrupole mass spectrometers, the FT ICR instrument is capable of resolving nominally isobaric components for identification based on elemental composition. Furthermore, FT ICR tandem mass spectrometry may be used as an additional aide to structure elucidation through the formation of structurally significant fragment ions. In addition during MS MS, FT ICR instruments provide high

5 MASS SPECTROMETRY FOR HIGH THROUGHPUT SCREENING 357 Figure 2. (a) Partial negative ion electrospray mass spectrum of a 36-component library mixture. Both the measured mass and the difference between the measured and theoretical values (in ppm) are shown. (b) Negative ion electrospray spectrum of the 120-component library showing the resolution of three nominally isobaric peaks. (Reprinted from Reference [16], 1998, with permission from Bentham Science Publishers.) resolution exact mass measurements of product ions. For example, the negative ion electrospray mass spectra of a 36 and a 120 compound peptide library mixture are shown in Figure 2. The resolution achieved in this experiment was In addition to peptide libraries, other types of combinatorial libraries may be analyzed as well using FT ICR mass spectrometers such as pyrazoles [13, 14] and pseudo-peptide macrocyclic hydrazones [15]. In the example from Fang et al. [16], the measured mass was within 10 ppm of the theoretical values. Although the exact masses of all components in a small combinatorial library can often be measured during a single infusion experiment, on-line HPLC separation or the analysis of discrete compounds is sometimes required to overcome ion suppression problems. Space-charge effects that cause poor resolution or low mass accuracy may be overcome by analyzing individual compounds or isolating and measuring narrow regions of the mass spectrum per analysis. Parallel LC MS Although mass spectrometry is inherently rapid and high throughput, LC MS is a relatively slow process due to the slow chromatographic separation step. Since LC MS is required in many instances for the analysis of mixtures and to eliminate interfering salts or buffers, two

6 358 Y.G. SHIN AND R.B. VAN BREEMEN Figure 3. (A) Parallel HPLC MS with four HPLC columns interfaced simultaneously to one mass spectrometer in order to increase the throughput of LC MS. In this example, eluent from each HPLC column is electrosprayed into a single ion source, but only one HPLC stream is sampled at a time by means of a rotating plate shown aligned with column B. (Based on Reference [18], 1999.) approaches have emerged to increase the throughput of this technique; parallel LC MS and fast LC MS. One approach to increasing throughput of the rate-limiting chromatographic separation has been to simultaneously interface multiple HPLC columns to a single mass spectrometer. This approach is called parallel LC MS. Commercial parallel electrospray interfaces and HPLC systems are now available that can accommodate upto 8 HPLC columns simultaneously [17 19]. Although the multiple sprays are introduced to the ion source simultaneously, these streams may be sampled in a timedependent manner to minimize cross contamination between channels. One approach to parallel spray interfaces consists of a multiple spray head assembly and a rotating plate that blocks all but one stream so that only one HPLC column is sampled at a time (Figure 3). Fast LC MS analysis The most popular approach to increasing the throughput of LC MS has been to minimize the time required for HPLC separation through an approach called fast HPLC. HPLC separations are accelerated by using shorter columns and higher mobile phase flow rates. In response to the growing demand for fast chromatography during LC MS, specialized products are being developed by HPLC column manufacturers. Since fast chromatographic separations are not carried out under ideal conditions and some coelution of species is likely to occur, the selectivity of the mass spectrometer or tandem mass spectrometer is essential for the characterization and/or quantitative analysis of the target compound. However, samples of compounds prepared using combinatorial chemistry are usually simple mixtures of reagents, by-products, and product that require only partial chromatographic purification to prevent ion suppression effects during mass spectrometric analysis. It should be noted that electrospray and APCI are the most widely used ionization methods for fast LC MS. In addition to molecular weight and structural confirmation (using MS MS) of synthetic products from combinatorial synthesis, fast LC MS is also used to assess the purity and yield of these compounds [8, 20]. Many researchers demand that combinatorial libraries be analyzed for both purity and structural identity prior to high throughput screening so as to assure the validity of structure-activity relationship that might be derived from the screening data. Since this approach requires that large numbers of compounds be analyzed for structure and purity prior to screening, high throughput analysis is essential to avoid creating a bottleneck in the drug discovery process. Fast LC MS and LC MS MS may be carried out using gradients (usually a step gradient with a reverse phase HPLC column) with a total cycle time of 1 3 min [21] or using an isocratic system requiring less

7 MASS SPECTROMETRY FOR HIGH THROUGHPUT SCREENING 359 than 1 min per analysis. A variety of HPLC columns are used for fast LC MS and include narrow bore (2 mm) and analytical bore (4.6 mm) diameter columns with column length typically from 3 to 5 cm. The mobile phase flow rate for the analytical columns is usually from 3 to 5 ml/min. Mass spectrometric identification of compounds in combinatorial libraries Two approaches are used most often in combinatorial synthesis: (1) parallel synthesis [22] and (2) split and pool synthesis [23]. The applicability of mass spectrometry to the analysis of combinatorial libraries is not limited to the analysis of synthetic products as a means of quality control, but also provides a valuable tool for the identification of active compounds during high throughput screening. Although the synthesis and screening of discrete compounds enables them to be followed through the entire process by using partial encoding or bar-coding, it is sometimes advantageous to screen library mixtures and use a technique such as mass spectrometry to rapidly identify the hit(s) in the mixture. One approach to the rapid deconvolution of combinatorial library mixtures is to prepare libraries containing compounds of unique molecular weight, and then identify them using mass spectrometry. However, such libraries are necessarily small since the molecular weight of most drug-like molecules is between 150 and 400 Da. Because of the molecular weight degeneracy of larger combinatorial libraries, several encoding strategies have been devised to rapidly identify active compounds in these mixtures [24 26]. Mass spectrometric deconvolution of ratio-encoded combinatorial libraries Since most combinatorial libraries contain compounds with degenerate molecular weights, various tagging strategies have been devised to uniquely identify library compounds bound to beads. Most of these tagging approaches are based on the synthesis of encoding molecules. For example, peptide [27] or oligonucleotide [28] labels have been synthesized on the beads in parallel to the target molecules and then sequenced for bead decoding. Alternatively, haloarene tags have been incorporated during synthesis and then identified with high sensitivity using electron-capture gas chromatography detection [29]. In addition to the increased time and cost for the synthesis of a library containing tagging moieties, the tagging groups themselves might interfere with screening giving false positive or negative results. For peptide libraries, one solution to this problem uses matrix-assisted laser desorption ionization (MALDI) mass spectrometry to directly desorb and identify peptides from beads that were screened and found to be hits [30]. This technique is called the termination synthesis approach. Since the peptide library compounds are analyzed directly, products with amino acid deletions or substitutions, side-reaction products, or incomplete deprotection are readily observed. Also, since there are no extra molecules used for chemical tagging, this source of interference is avoided. However, this approach is specific to peptide libraries and not necessarily applicable to other types of combinatorial libraries. Another approach that eliminates possible interference from the chemical tags, ratio encoding, has been developed for the mass spectrometric identification of bioactive leads using stable isotopes incorporated into the library compounds [26, 31]. Within the ligand itself, the code might be a single labeled atom that is conveniently inserted whenever a common reagent transfers at least one atom to the target compound or ligand. The code consists of an isotopic mixture having one of the many predetermined ratios of stable isotopes and can be incorporated in the linker or added through a reagent used during the synthesis. The mass spectrum of the compound shows a molecular ion with a unique isotope ratio that codes for a particular library compound. For example, Wagner et al. [26] synthesized a ratio-encoded 1000 member peptoid library using a mix and split approach. The four-step synthesis used a Knorr resin derivatized in the first step with DIC and a-bromoacetic acid encoded with a distinct ratio of 13 C and 12 C. For the ratio encoding, ten unique ratios were employed in 9% increments ranging from 9 : 91 to 90 : 10. The second position was encoded by the molecular weight of one of 10 different amine

8 360 Y.G. SHIN AND R.B. VAN BREEMEN Figure 4. Positive ion electrospray mass spectra of 2 isotope ratio-encoded peptoids each cleaved from a different single bead. (Reprinted from Reference [26], with permission from Bentham Science Publishers.) monomers added during the second step. The third and final amine monomer is known in the final pool. After the synthesis, a single bead could be cleaved, analyzed by mass spectrometry, and identified by the isotopic pattern and the molecular weight of the analyte peaks. For example, mass spectra of two peptoids cleaved from different beads are shown in Figure 4. Based on the isotope pattern and the molecular weight, each peptoid from the 1000 compound library could be identified. Since isotope ratio codes are contained within each combinatorial compound, a chemical tag is not required. This encoding technique is amenable to most chemical reactions and gives the added advantage of helping to identify products of side reactions. The speed of decoding allows the analysis of all active compounds from an assay and outperforms most other decoding technologies which are time consuming and decode a restricted set of active compounds. Therefore, potential structure activity relationship information may be more fully realized. Characterization and dereplication of natural products Although combinatorial synthesis provides rapid access to large numbers of compounds for screening during drug discovery and lead optimization, these libraries are usually based on a small number of common structures or scaffolds. There is a constant need for increasing the molecular diversity of combinatorial libraries and finding new scaffolds, and natural products have always been a rich source of chemical diversity for drug discovery. The traditional approach to screening natural products for drug leads utilizes bioassays to test organic solvent extracts for activity. If strong activity is detected, then activity-guided fractionation of the crude extract is used to isolate the active compound(s), which are identified using mass spectrometry (including tandem mass spectrometry and exact mass measurements), IR, UV/VIS spectrometry and NMR. Recently, a variety of mass spectrometry-based affinity screening methods have been developed to streamline the tedious process of activity-guided fractionation. These approaches are discussed in the next section. Whether lead compounds in natural product extracts are isolated using bioassay-guided fractionation or mass spectrometry-based screening, there is a high probability that the structure of the active compound(s) has already been reported in the natural product literature. In such cases, the tedious process of complete structure elucidation using a battery of spectrometric tools should be unnecessary. Instead, mass spectrometry alone may be used to quickly dereplicate or identify the known compounds based on molecular weight, fragmentation patterns and elemental composition in combination with natural product database searching [32 36]. Commercially available natural products databases include NAPRA- LERT [37], Scientific & Technical information Network (STN) [38], and the Dictionary of Natural Products [39]. Since some of these databases also contain UV/VIS absorbance data, it is also advantageous to use a photodiode array

9 MASS SPECTROMETRY FOR HIGH THROUGHPUT SCREENING 361 detector between the HPLC and mass spectrometer to obtain additional spectrometric data during LC UV MS dereplication [33, 34]. When using the traditional approach of bioassay-guided fractionation, the crude extract showing activity is fractionated using chromatography, and each fraction is re-assayed for activity. The throughput of this process has been enhanced by using HPLC with a 96- or 384-well plate fraction collector. Either the active fractions are re-chromatographed using LC UV MS for characterization and dereplication, or the entire eluate is analyzed using LC UV MS during fractionation. This latter approach was used by Cordell and Shin [33] for the dereplication of an extract of Begonia parviflora that showed cytotoxic activity against human oral epidermoid carcinoma (KB) in cell culture (Figure 5). The molecular weights of the compounds in the active fractions were compared to molecular weight data for compounds of Begonia sp. contained in the NAPRALERT database. As a result, cucurbitacin D, hexanorcucurbitacin D, cucurbitacin B, and dihydrocucurbitacin B were tentatively identified and then confirmed by demonstrating co-elution and identical electrospray mass spectra during LC MS. The use of mass spectrometry in the dereplication of natural products facilitates the rapid identification of active compounds for which Figure 5. (A) Extracted ion chromatogram of the active compounds in fractions of B. parviflora showing (B) cytotoxicity against human oral epidermoid carcinoma (KB). (Reprinted from Reference [33], 1999, with permission from IUPAC.)

10 362 Y.G. SHIN AND R.B. VAN BREEMEN structures have already been determined. As a result, time and effort are spared that might otherwise have been spent completely re-characterizing a previously determined structure. Furthermore, mass spectrometry, LC MS, and LC MS MS are extremely sensitive techniques requiring only trace amounts of the natural product (fmol pmol). Typically, complete spectroscopic characterization requires several orders of magnitude more material. Mass spectrometry-based screening In the previous sections, the utility of mass spectrometry was discussed for confirming the structures of compounds in combinatorial libraries, assessing the purity of library compounds, dereplicating natural products, deconvoluting combinatorial libraries, mass encoding libraries, and even for mass-triggered purification of library compounds. During the last few years, several mass spectrometry-based screening assays have been developed that are directed toward accelerating the pace of drug discovery and lead optimization. These methods are suitable for screening combinatorial library mixtures, and some are also useful for screening natural product extracts. All of the mass spectrometry-based screening methods use receptor binding of ligands as the basis for identification of lead compounds. Affinity chromatography mass spectrometry Since the introduction of affinity chromatography more than 30 years ago, this technique has become a standard biochemical tool for the isolation and identification of new binding partners to specific target molecules. Therefore, the coupling of affinity chromatography to mass spectrometry is a logical extension of this technique, and the application of affinity LC MS to the screening of combinatorial libraries has been demonstrated by several groups [40, 41]. During affinity LC MS screening, a receptor molecule such as a binding protein or enzyme is immobilized on a solid support within a chromatography column. The library mixture is pumped through the affinity column in a suitable binding buffer so that any ligands in the mixture with affinity for the receptor would be able to bind. Then, unbound material is washed away. Finally, the specifically bound ligands are eluted using a destabilizing mobile phase and identified using mass spectrometry. In some applications [40], ligands are eluted from the affinity column and then trapped on a second column such as a reverse phase HPLC column. LC MS or LC MS MS identification of the ligands (hits) is then carried out using the trapping column. In other systems, ligands are identified directly from the affinity column using mass spectrometry [41]. Usually, electrospray or APCI mass spectrometry are used, since they are the most robust and widely available interfaces for LC MS. For example, Kelly et al. [41] prepared an affinity column containing immobilized phosphatidylinositol-3-kinase and used it for direct LC MS screening of a 361-component peptide library. Electrospray mass spectrometry and tandem mass spectrometry were used to identify the ligands released from the affinity column using ph gradient elution. Advantages of affinity chromatography mass spectrometry for screening during drug discovery include versatility and re-use of the column. Both combinatorial libraries and natural product extracts can be screened using this approach, and a wide range of binding buffers may be used. Mass spectrometry-compatible mobile phases are only required during the final LC MS detection step. Furthermore, a single column may be used multiple times to screen different samples for ligands unless the destabilization solution irreversibly denatures, releases, or inhibits the receptor. Despite these advantages, affinity chromatography has numerous drawbacks that have prompted the development of alternative mass spectrometer screening tools. For example, immobilization of the receptor might change its affinity characteristics causing false negative or false positive hits. This is particularly problematic for receptors that are solution-phase in their native state. Also, developing and then implementing an immobilization scheme is often a slow, tedious and even expensive process, which is unique for each new receptor. Finally, false positive hits are often obtained when screening large, molecularly diverse libraries,

11 MASS SPECTROMETRY FOR HIGH THROUGHPUT SCREENING 363 since there are usually compounds in such mixtures that have affinity for the stationary phase or linker molecule instead of the receptor. Gel permeation chromatography (GPC) mass spectrometry Another type of chromatography that has been combined with mass spectrometry as a screening system for drug discovery is gel permeation chromatography (GPC) [42, 43]. Also called size exclusion chromatography, GPC separates molecules as they pass through a stationary phase containing particles with a defined pore size. Large molecules cannot enter the pores of the stationary phase and elute quickly from the GPC column. However, small molecules diffuse into the stationary phase and take longer to elute. During GPC-based screening, a library mixture is pre-incubated with a macromolecular receptor to allow any ligands in the library to bind. Then, GPC is used to separate the large receptor ligand complexes from the unbound compounds in the mixture. This separation step must be carried out quickly, since ligands begin to dissociate from the receptor immediately and can become lost during GPC. Despite this disadvantage, this approach allows both receptor and ligand to be screened in solution, which avoids some of the problems associated with the use of affinity columns for screening. Finally, the receptor ligand complex is denatured to release the ligands during chromatography on a reverse phase HPLC column, and the ligands are identified on-line using electrospray or APCI mass spectrometry. During the pre-incubation and GPC steps, any binding buffer may be used, since the binding buffer will be removed during reverse phase LC MS analysis. Also, any type of mass spectrometer may be used that is compatible with LC MS. In particular, high-resolution tandem mass spectrometers such as FT ICR instruments or hybrid QTOF mass spectrometers are ideal for the determination of the combinatorial libraries and identification of the hits. The GPC LC MS screening method should also be suitable for screening natural product extracts. Affinity capillary electrophoresis mass spectrometry Affinity capillary electrophoresis was originally used for the determination of the binding constants of small molecules to proteins [44 46]. This solution-based technique is rapid and requires very small amounts of analytes. Affinity constants are measured based on the mobility change of the ligand upon interaction with the receptor present in the electrophoretic buffer [47]. By combining affinity capillary electrophoresis with on-line mass spectrometric detection and identification, affinity constants for multiple compounds can be measured in a single analysis [6]. Recognizing that on-line mass spectrometric detection was helpful for the identification of each ligand, Chu et al. [48] extended this approach to include the screening of combinatorial libraries as a means of drug discovery. An example of affinity capillary mass spectrometry for the screening of a 100 tetrapeptide library for ligands to vancomycin is shown in Figure 6. Without vancomycin in the electrophoresis buffer, all the peptides eluted within 3 min. When vancomycin was present, the peptides eluted in order of affinity with the highest affinity compounds being detected between 4.5 and 5 min. Positive ion electrospray tandem mass spectrometry was used to identify the highest affinity ligands (Figure 6B). Note that the some peptide ligands such as Fmoc- DDFA were detected as adducts with Tris which was used as the electrophoresis buffer. Although the identification of this peptide was not prevented by the formation of this adduct with buffer, some buffers used in electrophoresis might interfere with mass spectrometric ionization and detection. Also, note that the types of libraries that have been screened using this approach have contained modest numbers of synthetic analogs such as peptides. Libraries exceeding 400 members required preliminary purification using affinity chromatography to reduce the number of compounds [48]. As a result, this approach is probably not ideal for screening large molecularly diverse libraries or natural product extracts. However, affinity capillary electrophoresis-mass spectrometry is fast

12 364 Y.G. SHIN AND R.B. VAN BREEMEN Figure 6. (A) Affinity capillary electrophoresis-uv analysis of a 100-tetrapeptide library, which elute in order of affinity for vancomycin (104 mm) included in the electrophoresis buffer. (B) Positive ion electrospray tandem mass spectrum with collision-induced dissociation of the Tris adduct of the protonated peptide showing the highest affinity for vancomycin and eluting at approximately 5 min in the electropherogram shown in (A). (Reprinted from Reference [48], 1996, with permission from the American Chemical Society.) with each analysis requiring 510 min. Also, this is a solution-phase screening method, and it may be used to measure affinity constants for ligand receptor interactions. Frontal affinity chromatography mass spectrometry For frontal affinity chromatography, receptor molecules are immobilized on a solid support packed into an affinity column [49], and ligands are continuously infused into the column. As compounds compete for binding sites on the affinity column, these sites become saturated until ligands begin to elute from the column at their infusion concentration. The compounds with the highest affinity for the immobilized receptor elute last, and compounds with no affinity elute immediately in the void volume. Therefore, this technique may be used to measure affinity constants for ligands in combinatorial libraries. Recently, frontal affinity chromatography was combined on-line with mass spectrometric detection and used for the screening of combinatorial libraries [50, 51]. In this application, mass spectrometry is used to continuously monitor the elution of all compounds in the library, and the last compounds to elute at their infusion concentrations represent the highest affinity compounds or hits. An example of the screening of 6 oligosaccharides with different binding affinities for an immobilized monoclonal carbohydrate-binding antibody is shown in Figure 7. Compounds 1 3 eluted immediately (no affinity) while compounds 4 6 eluted in order of increasing affinity for the antibody. Dissociation constants were determined to be 185, 12.6, and 1.8 mm for compounds 4 6, respectively [50]. Since frontal affinity chromatography utilizes a conventional affinity column, this technique provides additional applications of this type of column to investigators already using affinity mass spectrometry (see section on Affinity chromatography mass spectrometry). However, the same limitations and disadvantages of using immobilized receptors still apply such as nonspecific binding to the stationary phase, the development time and cost of preparing the affinity columns, and the possibility that immobilizing the receptor might alter its binding characteristics and specificity. In addition, mass spectrometric detection creates some additional limitations. Since all library compounds must be monitored simultaneously, the compounds must be selected so that they have unique molecular weights. Also, one compound in the mixture should not suppress the ionization of another. This is a potential problem that is difficult to predict and avoid. Finally, the binding buffer used for affinity chromatography must be compatible with on-line APCI or electrospray mass spectrometry. This means that the mobile phase must be volatile and usually of low ionic strength (i.e. typically 540 mm for electrospray ionization).

13 MASS SPECTROMETRY FOR HIGH THROUGHPUT SCREENING 365 Figure 7. Frontal affinity chromatography mass spectrometry for the screening of a six compound mixture. Top: Positive ion electrospray total ion chromatogram. Middle: Computer-reconstructed mass chromatogram of all six compounds. Compounds 1 3 break through simultaneously as indicated by the solid line. Bottom: Positive ion electrospray mass spectra recorded at times I, II, and III as indicated in the middle of trace. Protonated molecules corresponding to compounds 1 6 are labeled. At time I, signals for the non-ligands 1 3 are detected. By time III, all six compounds are detected. (Reprinted from Reference [50], with permission from Wiley-VCH Verlag GmbH.) Screening using electrospray Fourier transform ion cyclotron resonance (FT ICR) mass spectrometry Although FT ICR mass spectrometry is used routinely to determine the exact masses of library compounds and to confirm their structures using CID and high-resolution tandem mass spectrometry (see definitions of CID and MS MS in the Introduction), electrospray FT ICR mass spectrometry may be used for the direct screening of combinatorial libraries without the need for chromatography or ultrafiltration prior to analysis. In this application, a combinatorial library is pre-incubated with a receptor in solution and then analyzed directly using electrospray in order to identify receptor ligand complexes in the gas phase [52 56]. Once a receptor ligand complex is ionized and trapped in the FT ICR mass spectrometer, the mass difference between the complex and the receptor alone might be measured with sufficient resolution and accuracy to determine the mass(es) and perhaps elemental composition(s) of the ligand(s). If the ligand

14 366 Y.G. SHIN AND R.B. VAN BREEMEN carries a charge, then CID may be used to dissociate the ligand for subsequent analysis using tandem mass spectrometry. As an example, the FT ICR mass spectrometric screening of a 289-peptide library for ligands to carbonic anhydrase is shown in Figure 8 [53]. After electrospray ionization of a carbonic anhydrase ligand mixture, the receptor ligand complexes were isolated and trapped in the FT ICR mass spectrometer (Figure 8A). Then, CID and MS MS were used to dissociate and identify the peptide ligands (Figure 8B). The advantage of this screening method over other approaches is the elimination of purification steps prior to mass spectrometric identification. In addition to streamlining the screening process, Figure 8. (A) Negative ion electrospray FT ICR mass spectrometric screening of a 289-peptide library for ligands of carbonic anhydrase (CA-II, 2.5 mm) in a ph 7.0 buffer of 10 mm ammonium acetate. (B) Tandem mass spectrum with CID of the isolated complex of [CA-II+Zn+1] 9, where 1 represents the mixture of ligands with affinity for CA-II. The insets show expanded views of the singly charged inhibitors, [1] 1. The letters denote the amino acid compositions of the inhibitor ions. (Reprinted from Reference [53], 1996, with permission from the American Chemical Society.)

15 MASS SPECTROMETRY FOR HIGH THROUGHPUT SCREENING 367 the disadvantages associated with chromatographic separations are eliminated. An extension of this FT ICR mass spectrometry-based screening technique has been to screen a combinatorial library for ligands to two receptors simultaneously [55, 56]. In this example, the two receptors consisting of RNA constructs representing the prokaryotic (16 s) rrna and eukaryotic (18s) rrna A-site were incubated simultaneously with an aminoglycoside library to identify potential ligands. By screening a target mixture against the same library, screening efficiency is enhanced, and the number of analyses required is reduced. In spite of its many advantages, the use FT ICR mass spectrometric screening restricts the binding buffer and receptors that may be used. Only low ionic strength and volatile buffers are compatible with this approach (such as 10 mm ammonium acetate. See Figure 8). Also, the receptor and ligand must be highly purified to avoid impurities that might interfere with ionization and detection. Therefore, this technique is probably more suitable for the screening of combinatorial libraries than complex natural product mixtures. Finally, the receptor ligand complex must ionize efficiently during electrospray under solvent and ion source conditions that do not cause dissociation of the complex. Pulsed ultrafiltration mass spectrometry A versatile approach to screening solution phase combinatorial libraries and natural product extracts is pulsed ultrafltration mass spectrometry [57, 58], which utilizes a standard LC MS system with an ultrafiltration chamber substituted for the HPLC column. The principle of pulsed ultrafiltration screening of combinatorial libraries is shown in Figure 9. During pulsed ultrafiltration, ligand receptor complexes remain in solution in the ultrafiltration chamber while unbound library compounds and buffer are washed away. After unbound compounds are removed, the hits from the library are eluted from the chamber by destabilizing the ligand receptor complex using an organic solvent, a ph change, or a combination of both. The released ligands are identified on-line using APCI or electrospray mass spectrometry [57] or collected and analyzed off-line using mass spectrometry, LC-MS, or LC MS MS [59]. Figure 9. Scheme showing combinatorial library screening using pulsed ultrafiltration mass spectrometry. During the loading step (left), ligands are bound to the receptor either on-line (top) using a flow-through approach or off-line (bottom two incubations). Unbound compounds and binding buffer, cofactors, etc., are washed out of the ultrafiltration chamber to waste during a separation step (middle). Bound ligands are dissociated from the receptor molecules and eluted from the chamber by introducing a destabilizing solution such as methanol, ph change, etc. Finally, released ligands are identified using mass spectrometry, tandem mass spectrometry, or LC MS (right)

16 368 Y.G. SHIN AND R.B. VAN BREEMEN Figure 10. Identification of EHNA as the highest affinity ligand for adenosine deaminase in a combinatorial library of 20 adenosine analogs using ultrafiltration electrospray mass spectrometry. (Reprinted from Reference [57], 1997, with permission from the American Chemical Society.) An example of pulsed ultrafiltration mass spectrometry for the screening of a library of 20 adenosine analogs for ligands to adenosine deaminase is shown in Figure 10. After a 15 min preincubation of the library compounds (17.5 mm each except for EHNA, which was present at 1.75 mm) with 2.1 mm adenosine deaminase in 50 mm phosphate buffer, an aliquot containing 420 pmol of the receptor was injected into the ultrafiltration and washed for 8 min at 50 ml/min with water to remove the phosphate buffer and unbound or weakly binding library compounds. Methanol was introduced into the mobile phase to dissociate the enzyme-ligand complex and release bound ligands for identification by electrospray mass spectrometry. During methanol elution, only EHNA (erythro-9-(2-hydroxy-3- nonyl) adenine) was detected as the [M+H] + ion of m=z 278 (Figure 10). In control experiments using the library without enzyme, no library compounds were detected during methanol elution (Figure 10, Control). Despite being present at a 10-fold lower concentration than the natural substrate adenosine analogs, EHNA was easily identified since it had the highest affinity among the library compounds (K d =1.9 nm). This demonstrates the utility of ultrafiltration electrospray mass spectrometry for identifying a high affinity ligand among a set of analogs that bind to a specific receptor. In a follow-up lead optimization study using pulsed ultrafiltration mass spectrometry, a synthetic combinatorial library of EHNA analogs was screened for binding to adenosine deaminase, and structure-activity relationships for EHNA binding were identified [60]. As an illustration of the versatility of pulsed ultrafiltration mass spectrometry, binding assays for a variety of receptors have been reported including dihydrofolate reductase [59], cyclooxygenase-2 [58], serum albumin [61, 62] and estrogen receptors [63]. Not only is pulsed ultrafiltration useful for identifying ligands to different receptors, but a wide range of combinatorial libraries and natural product extracts in any suitable binding buffer may be screened. In addition to combinatorial libraries, complex natural product extracts have been screened [63], and neither plant nor fermentation broth matrices were found to interfere with screening [58]. As another example of the flexibility of this screening system, a centrifuge tube equipped with an ultrafiltration membrane [64] has been used instead of an on-line ultrafiltration chamber. Other applications of pulsed ultrafiltration mass spectrometry include screening drugs and drug candidates for metabolic stability [65], metabolic activation to reactive metabolites [66] and the measurement of affinity constants for ligand receptor interactions [61, 62]. Metabolism and toxicity screening applications of pulsed ultrafiltration use hepatic microsomes in the ultrafiltration chamber. For metabolic screening, drugs and the cofactor NADPH are flow-injected through the ultrafiltration chamber (oxygen is dissolved in the mobile phase), and the metabolites formed by microsomal cytochrome P450 and any unreacted compounds flow out of the chamber for mass spectrometric identification and/or quantitative analysis [65]. On-line applications require the use of volatile buffers, but LC MS and LC MS MS may be used off-line to analyze the ultrafiltrate no matter what buffer had been used. Screening drugs for metabolic activation using pulsed ultrafiltrationmass spectrometry is carried out in a similar manner, except that glutathione is coinjected along with NADPH and the drug substrate [66]. MS MS may be used on-line or LC MS MS used off-line to screen for glutathione adducts as an