Drug Discovery from Natural Products Using Affinity Selection-Mass Spectrometry

Abstract

As a starting point for drug discovery, affinity selection-mass spectrometry (AS-MS) is ideal for the discovery of lead compounds from chemically diverse sources such as botanical, fungal and microbial extracts. Based on binding interactions between macromolecular receptors and ligands of low molecular mass, AS-MS enables the rapid isolation of pharmacologically active small molecules from complex mixtures for mass spectrometric characterization and identification. Unlike conventional high-throughput screening, AS-MS requires no radiolabels, no UV or fluorescent chromophores, and is compatible with all classes of receptors, enzymes, incubation buffers, cofactors, and ligands. The most successful types of AS-MS include pulsed ultrafiltration (PUF) AS-MS, size exclusion chromatography (SEC) AS-MS, and magnetic microbead affinity selection screening (MagMASS), which differ in their approaches for separating the ligand-receptor complexes from the non-binding compounds in mixtures. After affinity isolation, the ligand(s) from the mixture are characterized using high resolution UHPLC-MS and tandem mass spectrometry. Based on these elemental composition and structural data, the identities of the lead compounds are determined by searching on-line databases for known natural products and by comparison with standards. The structures of novel natural products are determined using a combination of spectroscopic techniques including two-dimensional NMR and MS.

Graphical Abstract

1.1. Introduction

1.1.1. Role of affinity selection-mass spectrometry (AS-MS) in natural products drug discovery

Natural products are the most successful source of drugs and drug leads in the history of pharmacology [1], [2]. Although combinatorial chemistry currently receives more emphasis for lead discovery [3], Nature continues to be a source of unique chemical structural diversity for new drugs [4]. Approximately two-thirds of new small molecule drugs since 1981 have been natural products, derivatives of natural products, or mimics of natural products [5], and yet less than 10% of the world’s biodiversity has been evaluated for potential biological activity, so that many more useful natural lead compounds await discovery[1].

Currently the most widely used approach for natural products drug discovery, bioassay-guided fractionation uses iterative chromatographic fractionation followed by bioassay until an active compound is isolated [6]. In contrast, affinity selection-mass spectrometry (AS-MS) enables characterization and dereplication of active constituents in complex mixtures such as botanical extracts in a single experiment [7]. There are several variations of AS-MS, and the most successful have been pulsed ultrafiltration (PUF) ASMS [8], [9], size exclusion chromatography (SEC) AS-MS [10], [11], and magnetic microbead affinity selection mass spectrometry (MagMASS) [12] (Figure 1).

Figure 1.

Comparison of the affinity selection-mass spectrometry (AS-MS) approaches of pulsed ultrafiltration (PUF) AS-MS, size exclusion chromatography (SEC) AS-MS, and magnetic microbead affinity selection screening (MagMASS). Each approach shares the same incubation step during which equilibrium is established between ligands in a mixture of compounds and a macromolecular receptor, and the final analysis step using UHPLC-MS. These approaches differ in the separation step, during which ligands are affinity purified from the complex mixture. Also, PUF AS-MS and SEC AS-MS use solution-phase screening whereas MagMASS uses immobilized receptor.

Originally invented to screen pools of combinatorial libraries [9, 13], the van Breemen laboratory was among the first to recognize the applicability of AS-MS to natural products drug discovery [14]. Because AS-MS does not require advance knowledge of the identities of the compounds in a mixture, this screening technique is ideal for natural products. Whereas bioassay-guided fractionation can require days or weeks to isolate an active compound from an extract for characterization and identification, AS-MS can currently accomplish this entire process in just a few hours.

All AS-MS approaches begin by incubating a pharmacologically important receptor with a mixture of possible ligands (Figure 1). Then, the AS-MS approaches differ in how the ligand-receptor complexes are separated from non-binding molecules. Finally, all AS-MS approaches utilize UHPLC-MS to characterize the affinity-extracted ligands. The speed, selectivity and sensitivity of mass spectrometry with electrospray or atmospheric pressure chemical ionization make AS-MS ideal for discovering ligands to target receptors [15].

High resolution mass spectrometers are used for AS-MS so that elemental compositions may be determined from the hits in the chromatograms. This information is essential for the dereplication of hits from natural product sources. Data-dependent product ion tandem mass spectrometry may also be used to obtain additional structural information. Because the reversed phase UHPLC retention times of active compounds are known from the UHPLC-MS analysis, additional ligand may be purified for additional spectroscopic analysis as needed. In this way, novel natural product structures may be determined much faster than would be possible using bioassay-guided fractionation.

1.1.2. Advantages of AS-MS over conventional HTS

With less than 10% of the natural chemical diversity of the world explored to date, Nature offers an abundance of compounds for drug discovery. However, many of these compounds are fluorescent and interfere with conventional HTS methods that rely on detection techniques such as fluorescence polarization or FRET, and many more compounds that are natural products have strong UV/Vis chromophores that can interfere with spectroscopic absorbance-based HTS. Examples of fluorescent natural product classes include anthranilates, coumarins, alkaloids, polyenes, flavonoids, curcuminoids, polycyclic aromatic quinones, and azaphilones [16].

Based on mass spectrometric instead of fluorescence or absorbance detection, AS-MS is compatible with fluorescent and strong UV/Vis absorbing compounds (Table 1). Unlike HTS approaches requiring the use of radiolabeled ligands, AS-MS does not require radioisotopes. AS-MS is also compatible with all assay buffers and cofactors required for assaying pharmacological targets, and unlike HTS, AS-MS may be used to screen for ligands to new or understudied targets with unknown structure or function (Table 1) [17].

Table 1.

Features of AS-MS approaches and comparison with conventional HTS.

Feature	PUF-MS	SEC- ASMS	MagMASS	Conventional HTS
Combinatorial libraries	yes	yes	yes	yes
Compounds per assay	>2,000	>2,000	≤1,000,000	1 compound
Natural product mixture screening	yes	yes	yes	no
Consumption of reagents	low	low	low	high
Requires labeled reagents (radioisotopes, chromophores or fluorophores)	no	no	no	yes
Compatible with fluorescent and UV chromophores	yes	yes	yes	No Interference
Compatible with any assay buffer and cofactor	yes	yes	yes	no
Compatible with understudied targets	yes	yes	yes	no
Orthostatic ligand screening	yes	yes	yes	yes
Allosteric ligand screening	yes	yes	yes	no
Ranking of ligands by affinity to receptor	yes	yes	yes	multiple assays
Determination of affinity constants	yes	maybe	maybe	No
Solution-phase receptor screening	yes	yes	no	yes
Solid-phase receptor screening	no	no	yes	Yes
Allows receptor reuse	yes	no	yes	no

Another advantage of AS-MS over HTS is speed and cost. For example, MagMASS has been used screen peptide libraries containing up to 1 million compounds in a single assay (Table 1) [18], which is considerably faster and less expensive than testing one compound at a time in 1 million separate assays using HTS. The use of high-resolution MS and MS/MS during AS-MS provides elemental composition and structural information for lead compounds that would not be accessible using HTS to screen complex mixtures for ligands.

AS-MS provides additional capabilities that are not possible using conventional HTS such as the ability to rank order equimolar mixtures of ligands for affinity to a receptor in a single assay [19], [20]. or the ability to discover allosteric ligands [21], [22] (Table 1). Allosteric ligands bind to sites remote from the active site and can modulate conformation, ligand selectivity and affinity, and activity of the active site [23]. AS-MS may be used to distinguish orthosteric ligands from allosteric ligands by using competition assays involving a high affinity orthosteric ligand, but determination of the physical location of allosteric or cooperative binding requires separate structural studies such as x-ray crystallography or protein NMR. Because allosteric and orthosteric ligands bind to different sites on the same target, they can be detected simultaneously using AS-MS.

To determine if a lead compound is an allosteric or orthosteric ligand, a high affinity ligand is added to the incubation mixture, and the AS-MS assay is repeated. Lack of competition for the orthosteric site while still binding to the receptor indicates that the ligand binds allosterically. Follow-up experiments using techniques such as x-ray crystallography, proton NMR, or H/D exchange mass spectrometry may be used to determine where the allosteric ligand binds to the target.

2.1. Types of AS-MS

2.1.1. Pulsed ultrafiltration (PUF) AS-MS

One of the first AS-MS approaches, PUF AS-MS was invented in 1997 by the van Breemen laboratory [24]. PUF AS-MS screening begins with the incubation of mixture of compounds, such as a natural product extract (botanical, fungal or microbial), with a solution-phase macromolecular receptor (protein, enzyme, or RNA). After equilibrium is achieved, ultrafiltration is used to separate the large ligand-receptor complexes from the unbound low mass compounds (Figure 1A). Because large pore sizes enable faster ultrafiltration separation, the pore size of the ultrafiltration membrane should be as large as possible while still retaining the macromolecular receptor. The PUF step should be carried out quickly and at reduced temperature to minimize dissociation and loss of the ligands. Next, disruption of the receptor-ligand complexes using organic solvent and/or a pH change releases the ligands for UHPLC-MS analysis, while the macromolecular receptors are retained by the ultrafiltration membrane (Figure 1A). Typically, concentrations of ligand and receptor are in the low micromolar range for screening, which enables identification of ligands with dissociation constants in this range or better. A summary of the features of PUF AS-MS compared with SEC AS-MS, MagMASS and conventional HTS is shown in Table 1.

PUF AS-MS has been used to screen natural product mixtures for a variety of pharmacological targets (for a review, see Han et al. [25]) including the antiinflammation targets cyclooxygenase-1 [26], cyclooxygenase-2 [27], secretory phospholipase A2 [28], and 5-lipoxygenase [29], and 15-lipoxygenase. Cancer targets have included retinoid X receptor-α [30], and urokinase-type plasminogen activator [31], and Parkinson’s disease and Alzheimer’s disease targets have included tyrosinase [32] and acetylcholinesterase [33]. Antibiotic and anti-malarial targets for PUF AS-MS have included quinone reductase-2 [34], eukaryotic dihydrofolate reductase [35], Plasmodium falciparum thioredoxin and glutathione reductases [36], and anti-viral applications have targeted neuraminidase [37], and Ebola and Marburg virus nucleoproteins [38]. Not limited to protein or RNA targets, PUF AS-MS has been used to screen natural products for ligands to mitochondria [39].

2.1.2. Size exclusion chromatography (SEC) AS-MS

Invented by Kaur, et al., [40] SEC AS-MS is a solution-phase screening approach like PUF AS-MS (Table 1) that begins with the incubation of a mixture of possible ligands with a macromolecular receptor (Figure 1B). After equilibrium is achieved, SEC is used to separate the large ligand-receptor complexes from smaller, unbound compounds. The high mass complexes elute first during SEC and are then denatured using organic solvent to release the ligands for reversed phase LC-MS analysis. Note that the size exclusion column may be connected directly to the reversed phase column in a 2D-chromatography configuration as described by Blom, et al., [41] and then modified by Annis, et al., [42]

SEC AS-MS with 2D-chromatography is becoming popular in the pharmaceutical industry for the discovery of drug leads [43]. Although these application have been focused on combinatorial library screening, targets have been as diverse as bacterial RNA polymerase [43], ncRNA bacterial riboswitches [44], and the Myc oncogene that binds to the G-quadruplex [45]. In one of the few natural product applications, SEC AS-MS was used to screen a mixture of isomeric panafungins from Fusarium larvarum for ligands to the anti-fungal target polyadenosine polymerase [46].

2.1.3. Magnetic microbead affinity selection screening (MagMASS)

Providing the same advantages over conventional HTS as PUF AS-MS and SEC AS-MS (Table 1), MagMASS was invented in the van Breemen laboratory [12] as a solid-phase alternative to complement these solution-phase screening approaches. MagMASS involves tethering the target to magnetic microbeads, incubating the immobilized protein with a natural product mixture, using magnetism to separate the ligand-protein/bead complexes from unbound compounds, and then releasing the bound ligands for UHPLC-MS analysis (Figure 1C). Like all AS-MS methods, the separation of bound from free compounds needs to be carried out quickly and at reduced temperature (usually 4°C) to minimize ligand loss due to premature dissociation from the receptor. However, due to faster separation, MagMASS offers superior sensitivity compared with other AS-MS approaches.

Immobilization of receptors for screening is preferred when they are unstable in solution (such as proteases that can undergo autoproteolysis) or when they can be immobilized in conformations analogous to their native state (such as receptors on the surface of cells, cell organelles, or virus particles). For MagMASS, the pharmacological target may be immobilized covalently or non-covalently. Covalent immobilization is carried out using magnetic beads that are chemically modified to introduce nucleophilic or electrophilic functional groups that react with crosslinkers or directly with functional groups on the receptor. Examples of covalent immobilization approaches include reaction of receptor primary amino groups with maleimide or N-hydroxysuccinimide on the surface of magnetic beads or reaction of primary amino groups on the beads with carboxylic acid groups on the receptor following activation to electrophilic esters using a water soluble diimide such as 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide [47]. Non-covalent immobilization takes advantage of an affinity tag added at a specific position in the receptor that binds strongly to a complementary affinity group on the surface of the magnetic microbead. Examples of non-covalent immobilization between the receptor tag and functionalized magnetic microbead include biotin and streptavidin, lectins and sugars, maltose binding protein and amylose, histidine and nickel ions, and glutathione S-transferase and glutathione. Immobilization of the receptors in specific orientations can help preserve their activity. Natural product applications of MagMASS have included targets such as 15-lipoxygenase [48], monoamine oxidase A [49], and the 5-HT2C G protein-coupled receptor [50].

2.2. Automation and Lead Compound Identification

To expedite screening using AS-MS, multiple ligand mixtures can be incubated in parallel using a microwell plate and processed using robotic liquid handling systems [48]. Whether using PUF AS-MS, SEC LC-MS or MagMASS, the rate limiting step in this process is the UHPLC-MS analysis. An internal standard may be added to each sample immediately before analysis so that each chromatogram may be normalized, and then metabolomics software (such as XCMS online) [51] used to compare the UHPLC-MS chromatogram of each sample with the negative control chromatograms to find unique peaks corresponding to affinity selected ligands to the target protein.

The identification of ligands from natural product extracts begins with high resolution MS and MS/MS analyses to determine elemental composition and structural features. Natural product lead compounds for which structures have been reported in the literature can be identified based on their elemental compositions, tandem mass spectra, and comparison with standards in a process called dereplication. Reversed-phase UHPLC retention times (obtained during screening step) enable mass-directed LC-MS purification to isolate novel ligands [52] for complete structure determination. Novel natural products usually require analysis using a combination of spectroscopic techniques for structure determination including high-resolution mass spectrometry, 1D and 2D NMR (COSY, TOCSY, HMQC, HSQC, HMBC, and NOESY), etc.

3.1. Conclusions

During the first two decades of development, PUF AS-MS, SEC AS-MS, and MagMASS have become faster, more robust and automated. Faster and less expensive than conventional HTS, AS-MS requires no fluorescent, UV-active, luminescent, or radioactive reagents. Also, the range of AS-MS applications has grown to include many targets that have resisted conventional HTS implementation. Whereas conventional HTS is designed to discover only orthosteric ligands, AS-MS also enables the discovery of allosteric ligands.

Due to compatibility with mixtures as well as with compounds exhibiting fluorescence or strong UV chromophores, AS-MS facilitates the screening of natural products in extracts of botanicals, fungi, and microbial cultures. Therefore, AS-MS provides access to compounds with much greater chemical diversity than is available from synthetic chemical libraries. As AS-MS matures and is more widely adopted during the next two decades, expect AS-MS to become the most productive approach to generate lead compounds for drug discovery.

Highlights.

• AS-MS assays enable the discovery of lead compounds from complex mixtures

• Suitable for screening natural products in botanical, microbial and fungal extracts

• Compatible with all ligands and targets, including those unsuitable for HTS

• AS-MS requires no radiolabels, fluorescent tags or strong chromophores

Abbreviations

AS-MS

affinity selection-mass spectrometry

HTS

high-throughput screening

MagMASS

magnetic microbead affinity selection screening

PUF AS-MS

pulsed ultrafiltration AS-MS

SEC AS-MS

size exclusion chromatography AS-MS