The future of drug discovery: enabling technologies for enhancing lead characterization and profiling therapeutic potential

Abstract

Technology often serves as a handmaiden and catalyst of invention. The discovery of safe, effective medications depends critically upon experimental approaches capable of providing high-impact information on the biological effects of drug candidates early in the discovery pipeline. This information can enable reliable lead identification, pharmacological compound differentiation and successful translation of research output into clinically useful therapeutics. The shallow preclinical profiling of candidate compounds promulgates a minimalistic understanding of their biological effects and undermines the level of value creation necessary for finding quality leads worth moving forward within the development pipeline with efficiency and prognostic reliability sufficient to help remediate the current pharma-industry productivity drought. Three specific technologies discussed herein, in addition to experimental areas intimately associated with contemporary drug discovery, appear to hold particular promise for strengthening the preclinical valuation of drug candidates by deepening lead characterization. These are: i) hydrogen–deuterium exchange mass spectrometry for characterizing structural and ligand-interaction dynamics of disease-relevant proteins; ii) activity-based chemoproteomics for profiling the functional diversity of mammalian proteomes; and iii) nuclease-mediated precision gene editing for developing more translatable cellular and in vivo models of human diseases. When applied in an informed manner congruent with the clinical understanding of disease processes, technologies such as these that span levels of biological organization can serve as valuable enablers of drug discovery and potentially contribute to reducing the current, unacceptably high rates of compound clinical failure.

Keywords::

• activity-based protein profiling
• disease models
• drug targets
• enzymes
• functional proteomics
• genome editing
• lead profiling
• ligand interaction
• molecular pharmacology
• nucleases
• protein dynamics
• receptors
• structural biology

1. Technology informs and enables translational drug innovation

Multiple challenges blanket the pharmaceutical industry under darkening clouds of inefficiency and instability as staggering development costs and moribund drug-approval rates erode returns on the immense up-front time and capital investments required for therapeutics invention [1-3]. Many unsolved medical problems persist for which safe, effective, population-based therapies are lacking, particularly with regard to complex, multifactorial diseases whose etiologies and therapeutic outcomes involve dynamic interplay among myriad environmental and genetic factors and biological signaling pathways, often across multiple organs/organ systems [4]. Pharma-industry’s challenges and opportunities have made the identification and characterization of agents that produce a desired salutary effect on a disease process a central focus of preclinical drug discovery in order to triage and optimize those very rare advanced leads worthy of human testing. The view has been espoused that shallow preclinical profiling of drug leads has decisively undermined sustainable growth potential in the pharmaceutical industry across therapeutic areas and indications. Myopic appreciation of the biological properties of lead drug candidates, in turn, can jeopardize product innovation and limit value creation by sanctioning leads of insufficient quality, validity and translational promise for pipeline advancement and entry into clinical trials [5]. Inadequate preclinical lead characterization is, therefore, a likely contributor to the unprecedentedly high failure rates of investigational drugs despite the increasing numbers of compounds in development and burgeoning total research and development (R&D) expenditures [6].

These circumstances suggest the need to raise the quality standards for development leads and strengthen lead-validation packages through more comprehensive and disease-relevant biological profiling. Although this view may appear to position medicinal chemistry ancillary to pharmacologial drug-discovery endeavors, the author considers that synthetic chemistry is operationally better defined and standardized than is biological profiling, the discovery ideal represented by coherence between medicinal chemistry and biological/pharmacological research as informed by the clinic. Arguably, then, a major – if not decisive – value-added proposition in preclinical drug discovery comes from thorough and rigorous compound characterization in disease-relevant biological systems, information from which should iteratively inform allied medicinal chemistry efforts. Toward this end, technological advances affording more comprehensive, in-depth candidate characterization have been demonstrated to empower preclinical discovery efforts [7].

From this vantage point and the author’s R&D experience of over three decades in discovery-related activities within both public and private sectors that generated numerous development leads and clinical candidates as well as approved therapeutics for diverse indications, this commentary will discuss three specific technologies that, in the author’s opinion, have demonstrated promise for enriching the innovation economy of drug development by improving compound profiling and enhancing our understanding of disease mechanisms. Since the technologies considered are themselves the subjects of refinement and scope and application extensions, their ultimate impact on therapeutics innovation cannot be definitively gauged at present. The primary purpose of this presentation, therefore, is to exemplify and promote awareness of emerging technologies for their potential to enrich the preclinical profiling of drug leads and strengthen lead validation in ways directly relevant to evaluating therapeutically attractive drug candidates. Small-molecule discovery aimed at modulating protein (enzyme, receptor) function will be emphasized, since research along this line has generated the majority of known therapeutics and is foundational to most contemporary discovery efforts [1,3].

2. Dynamic structural biology of disease-related proteins: hydrogen–deuterium exchange mass spectrometry

The 2012 Chemistry Nobel Prizes well illustrate the appreciable research and translational significance of ‘chemical pharmacology,’ especially structure–function correlates of protein modulation by therapeutic ligands [8]. Three-dimensional crystal structures of disease-relevant proteins have aided understanding of their interaction profiles with small-molecule drugs, thereby informing structure-led drug design [9]. Recent methodological advances in (especially membrane) protein isolation and crystallization, X-ray sources and allied instrumentation, and data handling auger well for future crystallographic protein analyses [10]. Nonetheless, unease exists as to the fidelity with which the conformation of an isolated, crystalline protein is able to represent that protein’s native structure and the degree to which even a suite of non-contradictory static X-ray maps per se can reflect the diversity of a protein’s conformational repertoire [11]. The concern is underscored by the important roles that protein structural transitions play in the therapeutic action of drugs on receptor-mediated information transmission and enzyme catalysis [12,13].

During the past two decades, the technique of peptide-level hydrogen–deuterium exchange (HDX) mass spectrometry (MS) (HDX-MS) has met with increasing application in the experimental interrogation of higher-order protein conformation and structural dynamics. As detailed in comprehensive historical and technical reviews [14-16] and schematized (Figure 1), HDX-MS leverages two phenomena: i) the unique, fundamental property of protein backbone amide hydrogen atoms to exchange with solvent deuterons from ‘heavy’ water (i.e., D₂O) in real time under physiological conditions; and ii) the ability of MS to quantify the protein mass so increased. Under set conditions of pH and temperature, the pattern of hydrogen exchange reflects the degree of solvent accessibility and intramolecular hydrogen bonding in a conformationally and structurally sensitive manner. Generally, solvent-exposed and highly dynamic protein regions (e.g., unstructured domains interposed between helices or loops) will exchange rapidly. Domains involved in hydrogen-bonding networks or within the protein’s interior (e.g., buried α-helices or β-sheets) tend to be less dynamic and exchange more slowly. Alteration in the location and/or rate of deuterium incorporation signals a motional change in solvent accessibility and/or hydrogen bonding in the protein. The basic HDX-MS technology may be augmented with sophisticated MS techniques to enhance spatial (i.e., from short peptide regions of the protein [Figure 1] to the single amino acid level) and temporal (i.e., in the sub-second time frame) resolution of the deuteration profile [17-19].

Figure 1. Diagrammatic illustration of prototypical HDX-MS analyses. Under physiological conditions (temperature, pH), a protein sample in compatible buffer is diluted with the same buffer containing 99.9% ‘heavy’ water (D₂O) in place of H₂O. Deuteration of the intact protein is allowed to proceed for various durations prior to quenching by rapid acidification to pH 2.5 and cooling to 0°C. The intact deuterated protein may either be directly injected into the mass spectrometer or digested under quench conditions with an acid protease prior to UPLC of the resulting peptide hydrolysate at 0°C (to restrict deuterium back exchange) followed by peptide-level mass analysis by MS. The mass spectra allow determination of the kinetics of deuterium uptake by the protein and/or its peptic peptides, the latter identifiable by MS/MS of the protein itself.

HDX-MS has enabled experimenter access to details on the intrinsic structures and conformational dynamics of a range of proteins and the functional implications of their structural properties/dynamics [16,20,21]. Recent specific examples of discovery-related HDX-MS demonstrate that HDX-MS can afford unique insights into the function and dynamics of disease-relevant protein targets and protein therapeutics unobtainable by other structural biology techniques (Table 1) [22-31]. In these regards, HDX-MS was key to obtaining the first definitive experimental corroboration of implications from molecular-dynamics simulations and energy landscape theory that local unfolding guides intermediate structural trajectories among active and inactive receptor conformations. Existing crystal structures of the druggable protein studied (the EGFR) did not correlate with the simulated intermediate protein conformations [21]. HDX-MS enabled definition of structural elements essential to selective therapeutic-ligand recognition by pathologically relevant proteins and informed thereby the design and optimization of novel enzyme and receptor modulators with improved potency, selectivity and/or functional specificity – properties often decisive to the pipeline viability of new chemical entities as leads toward clinical candidacy. Regional changes in DOT1-like, histone H3 methyltransferase upon binding of a potent, highly selective inhibitor apparent by HDX-MS, but not from crystals of the liganded enzyme, informed medicinal chemistry efforts by suggesting specific chemical modifications that improved the inhibitor’s potency and pharmacokinetics to realize a potential therapeutic [24]. Given the mounting interest in functionally selective or ‘biased’ drugs whose molecular pharmacology is capable of routing information output away from signaling pathways that elicit adverse events and toward those that are most therapeutically relevant [32], HDX-MS has been used to identify and distinguish ligand-induced receptor conformational responses associated with distinct molecular pharmacologies (e.g., receptor agonist vs antagonist vs inverse agonist) [28,29,33] and modes of ligand engagement (e.g., covalent vs noncovalent enzyme inhibitors) [22] that are therapeutically beneficial. In these studies [22,28,29,33], HDX-MS analyses enabled definition of the molecular determinants of the interactions between ligands as candidate drugs and therapeutic targets, data that helped tune the resulting information output to salutary ends. This information positively impacted the design of novel ligands that modulate in a therapeutically beneficial manner the function of key receptors and enzymes associated with diseases representing significant unmet medical needs and major unsolved health problems (Table 1).

Table 1. Exemplary discovery-related hydrogen–deuterium exchange mass spectrometry studies focused on disease-related proteins or protein therapeutics.

Protein	Therapeutic indication(s)	Information obtained	Ref.
Enzymes
Monoacylglycerol lipase	Inflammation, pain, CNS injury, cancer, anxiety disorders	Distinctive structural responses to reversible vs irreversible inhibitors	[22]
		Effect of membrane association on inhibitor engagement	[23]
DOT1L H3 MT	Cancer	Inhibitor-associated active-site remodeling/adaptation	[24]
K-Ras	Cancer	Selective inhibitor targeting	[25]
ClpP protease	Antibacterial	Identification of inhibitor binding epitopes	[26]
Receptors
Retinoid X	Cancer	Agonist-induced helix dynamics responsible for initiating signaling	[27]
AMPA	Schizophrenia, depression, neurodegenerative diseases	Differentiation of binding modes between full vs partial agonist	[28]
β2-adrenergic	Asthma, COPD	Differentiation of interaction profiles among biased agonists	[29]
Notch3	Cancer, chemotherapy resistance	Modulatory antibody dynamics	[30]
Protein therapeutics
IgG1 recombinant mAb	Allergy, immunodeficiency, cancer, autoimmunity	Structural mechanism of instability/aggregation	[31]

AMPA: α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid; ClpP: Caseinolytic protease; COPD: Chronic obstructive pulmonary disease; DOT1L H3 MT: DOT1-like, histone H3 methyltransferase; IgG1: Immunoglobulin G1; K-Ras: Kirsten rat sarcoma viral oncogene homolog.

3. Functional annotation of mammalian proteomes: activity-based enzyme chemoproteomics

Enzymes catalyze discrete biological reactions that allow molecular transformations to take place within time frames compatible with the physiological activities and homeostasis essential to living systems. Modulation of enzyme function is a principal mode of (especially small-molecule) drug action, making the identification of pathologically significant enzymes and validation of their therapeutic relevance prominent, long-standing concerns of drug discovery and development [34]. These considerations lend great importance to the annotation of enzyme function in health and disease and definition of the interaction specificity or promiscuity of designer small-molecule modulators with enzymes, the significance of which is likely to intensify, given our deepening appreciation of the complexities that underlie enzymatic regulation of metabolic pathways and information networks key to disease etiology and treatment [35]. Specifically, detailed knowledge of the interaction spectrum of a potential drug candidate within a cellular/tissue proteome would be of great value in uncovering unwanted off-target interactions and as indicator of target/signaling pathway selectivity, information that can help guide medicinal chemistry efforts for optimizing these properties in a lead compound to therapeutic ends.

For these purposes, activity-based protein profiling (ABPP) is being embraced as a proteomic platform for chemically interrogating and quantifying the functional state of enzyme families in native biological samples. As detailed in reviews [36-38] and illustrated in Figure 2, prototypic ABPP technology utilizes active site-directed covalent probes that target a specific enzyme subset in a cell/tissue proteome characterized by shared mechanistic features underlying catalytic activity. Probe design incorporates a reactive ‘warhead’ (e.g., an electrophilic or photoreactive group) to label covalently active enzymes linked to an analytical (e.g., fluorophore or affinity) tag for detection and/or enrichment of the probe-labeled enzymes. The warhead binding group is designed to direct the probe toward select enzyme classes. Probe design allows for wide variations in the stringency of probe specificity among enzymes that share a similar catalytic mechanism and/or substrate selectivity. The resultant, probe-labeled proteome may then be analyzed in various ways. After affinity purification by exploiting a suitable analytical tag on the probe (e.g., with avidin beads for a biotin-functionalized probe and on-bead digestion), the targets of ABPP probe engagement can be quantified and identified by multidimensional liquid chromatography–tandem mass spectrometry (LC–MS/MS) with MS protein-sequence algorithm software. Alternatively, the probe-labeled proteome can be digested, and the labeled peptides affinity purified, cleaved from the affinity matrix and subjected to LC–MS/MS to identify directly the sites of probe labeling and target engagement in the tagged enzyme(s). Another ABPP variant, ‘competitive ABPP,’ can be used to screen for enzyme inhibitors by incubating the proteome under study with a putative inhibitor prior to introducing a fluorescent ABPP probe. Comparison by PAGE with a parallel sample reacted with the ABPP probe allows for quantitative in-gel activity analysis by virtue of the reduction in fluorescence band intensity at the inhibited enzyme(s), affording rapid identification of the inhibitor’s target(s) [and off-targets(s)] (Figure 2).

Figure 2. Schematic outline of prototypical ABPP enzyme analyses. A cell/tissue proteome is incubated with a chemical probe () bearing both a warhead designed to react covalently with catalytically active enzymes in a specific enzyme/enzyme class () and a fluorophore or affinity moiety (). The probe-labeled proteome can then be subjected to various analyses, including: (A) isolation of the probe-labeled proteins on affinity beads followed by on-bead enzymatic digestion and LC–MS/MS analysis to identify the modified proteins; (B) tryptic digestion of the labeled proteome followed by affinity purification, enzymatic digestion or cleavage, and LC–MS/MS to identify the sites of probe modification/target engagement at the amino acid level; (C) ‘competitive ABPP’ involving electrophoretic resolution of the proteome on polyacrylamide gels and in-gel fluorophore detection of the labeled proteins (red bands). Prior incubation of the proteome with a putative enzyme inhibitor (‘with inhibitor’) will block the inhibited enzyme’s reaction with the ABPP probe and obviate fluorescence in that band as compared to the proteome sample not previously treated with putative enzyme inhibitor (‘no inhibitor’), as indicated by the *symbol to the right of the electrophoretograms, affording profiling of the inhibitor’s target and selectivity within the proteome studied.

ABPP technology has allowed initial functional and mechanistic characterization of enzymes, elucidation of the roles of enzymes in metabolic control and cell-signaling information pathways, and definition of the influence of natural products/metabolites on enzyme activity [39-44]. With regard to drug discovery, ABPP aided identification of several druggable enzymes of pathological relevance and development of novel, therapeutically promising enzyme inhibitors: Competitive ABPP played a critical role in demonstrating the involvement of monoacylglycerol lipase (MGL) in aggressive cancer and designing novel, selective MGL inhibitors that show preclinical efficacy against oncological end points in vitro and in vivo [45,46]. Although excessive myeloperoxidase (MPO) activity has been implicated in the etiology of several diseases (e.g., rheumatoid arthritis, atherosclerosis, Parkinson’s disease), druggable MPO inhibitors have proven elusive. ABPP recently enabled the first identification of a mechanism-based MPO inhibitor with sufficient selectivity to serve as a tool compound for studies of the role of MPO in disease and as a template for rational drug design [47]. Competitive ABPP proved instrumental in identifying novel carboxylesterase 3 (Ces3) inhibitors effective preclinically against metabolic syndrome in vivo [48], thereby substantiating murine gene-modification data implicating Ces3 as a driver of pathogenic lipid accumulation associated with obesity and diabetes [49] and pointing the way to pharmacological inhibition of the mouse enzyme’s human ortholog, Ces1, as a therapeutic modality for metabolic disease. Perhaps of greater overall significance for enabling drug discovery, ABPP applied to a preliminary cell-based (adipocyte) screen of a compound library allowed unambiguous identification of Ces3 as the enzyme whose inhibition was responsible for the salutary metabolic effects observed [48], demonstrating a clear niche for ABPP in bridging target-based and phenotypic discovery approaches for annotating and deconvoluting disease-causing enzymes, finding directed therapeutic modulators with high confidence, and illuminating disease mechanisms.

4. Modeling complex human diseases: precision genome editor technologies

Living systems that display phenotypic features of a human disease have made uncontestable contributions to new therapeutics development. Yet cellular and rodent disease models conventionally used for drug discovery are intrinsically limited in scope and relevance. The range of standard immortalized cell lines and human cells able to be sampled noninvasively and studied in primary culture hardly begins to approach the variety of cell types involved in human disease. Likewise, decades of pharmacological data from murine disease models (commonly, single-gene-disruption ‘knock-out’ animals) have underscored the myriad metabolic, biochemical and physiological differences between rodents and humans that render such systems far from translationally optimal [50]. The prognostic value of the current stock of cell- and animal-based systems has been considered sufficiently tenuous to engender increasingly noisome concern regarding the reliability with which results from extant disease models may be translated to humans [51]. This situation invites improved experimental disease models for preclinical drug profiling that allow extrapolation to the clinic with heightened confidence for predicting the clinical outcome and human impact of potential diagnostic stratagems and therapies [52]. In particular, living systems that display multiple contributors to disease etiology rather than one single characteristic of a human-disease phenotype should be most useful for deepening our understanding of mechanisms underlying therapeutic activity, drug–drug interactions and adverse-event profiles.

Beyond attempts at deriving curative, gene- and cell-based therapies and augmenting production-agriculture yields [53,54], precision genome editing (PGE) has greatly enhanced the prospect of improved, tailor-made models of (especially multifactorial) human diseases. Nuclease-mediated PGE technology enables targeted introduction of discrete, single or multiple disease-related gene modifications into the chromosomal DNA of human cells and laboratory animals (including nonhuman primates), endowing these living systems with disease-specific signatures [54,55]. The historical and technological details of genome modification through nuclease-mediated PGE are available in print overviews [56-59] and on video (http://www.youtube.com/watch?v=zDkUFzZoQAs). Essentially, the most popular nuclease-mediated PGE technologies utilize one of three gene-editing enzymes – zinc-finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs) or clustered regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated (Cas) nuclease-based RNA-guided DNA endonucleases (also known as the type-II CRISPR system) – as ‘molecular scissors.’ The resulting double-strand DNA breaks recruit endogenous DNA damage-repair mechanisms that either induce local mutations or stimulate homologous recombination with exogenously provided donor DNA. Site-specificity in the ZFN and TALEN systems relies upon custom-made, programmable DNA-binding protein domains, whereas the CRISPR-Cas9 system attains specificity by virtue of an engineered RNA that guides the endonuclease. These PGE platforms enable custom, predetermined DNA sequence changes within natural chromosomal contexts at defined genomic sites as well as transgene addition at specific genomic loci virtually anywhere in complex genomes with a level of control allowing introduction of both single and multiple disease-relevant point mutations.

As demonstrated in several proof-of-principle studies discussed elsewhere [59-62], PGE nuclease tools have been used to introduce targeted gene modifications into diverse cell types (including primary cells, [transformed human] cell lines, and adult or embryonic human stem cells) and animal species routinely (e.g., rat, mouse, hamster, zebra fish) and less frequently (e.g., pigs, monkeys, marmosets) employed in laboratory research. With respect to pathologically relevant cell systems, nuclease-based PGE has opened revolutionary possibilities in hereditable cellular reprogramming for introducing discrete disease mutations into (human) cells that can then be differentiated in culture to display aberrant phenotypic manifestations indicative of a specific disease [63-72]. Particularly valuable in (patient-specific) disease situations where relevant human tissue sampling/in vivo experimentation is impractical or ethically questionable (e.g., many neurodegenerative syndromes [63]), the resultant cell-based ‘disease-/patient-in-a-dish’ systems have already helped inform drug discovery in several ways: screening for and profiling (potential) therapies and mechanisms of drug resistance [65,67,70,71]; selecting and monitoring pathologically relevant therapeutic targets [63-65,67]; answering mechanistic questions of disease pathophysiology [63-65,69]; and determining the effects of genetic contributions and drug sensitivities on disease phenotypes/human genetic disorders [63,66,68] (Table 2). Likewise, mutations introduced into various laboratory animals by nuclease-mediated PGE have enabled construction of in vivo models of human diseases involving single or multiple mutations that have been used for key discovery-related activities, including profiling of potential therapeutics, identification of the genetic basis of disease and definition of the causal relationships between diseases and human gene variants identified in genome-wide association studies (Table 2) [73-80]. This last application of nuclease-mediated PGE is particularly intriguing for therapeutics invention, since it represents a potentially more relevant and informative ‘bedside-to-bench’ approach distinct from the typical, marketing-driven ‘bench-to-bedside’ discovery route.

Table 2. Exemplary discovery-related cell and animal systems generated through nuclease-mediated precision genome editing.

System	PGE technology	Discovery application	Ref.
Cell-based
Dopaminergic neurons with LRRK2 mutation	ZFN	Genetic contributors to Parkinson’s disease phenotypes	[64]
WAS knockout K562 human myelogenous leukemia line	ZFN	Megakaryocyte differentiation; Wiskott–Aldrich syndrome model	[65]
APP knock-in fibroblasts	ZFN	Anti-Alzheimer disease drug screen	[66]
HPRT1 knockout neurons from human iPS	TALEN	Lesch–Nyhan syndrome model	[67]
Reporter K562 human myelogenous lines	TALEN	Globin-modulator screen for sickle cell disease	[68]
Human colorectal cell line	TALEN	Genetic basis of PCS(MVA) syndrome	[69]
DNMT3B double-allele human iPS mutants	CRISPR	ICF syndrome model	[70]
Knockouts in A375 human melanoma line	CRISPR	Genetic basis of cancer-drug resistance	[71]
Knockouts in human kidney and leukemia lines	CRISPR	Genome-wide functional screening	[71,72]
Animal
Hepatic F9 knockout mice	ZFN	Hemophilia model	[73]
IL2RG knockout pigs	ZFN	XSCID model	[74]
Tlr4 knockout rats	TALEN	Ethanol action/inflammatory disorder model	[75]
LDLR knockout swine	TALEN	Familial hypercholesterolemia model	[76]
Rab38 mis- and nonsense mouse mutants	TALEN	Hermansky–Pudlak syndrome model	[77]
DAZL or APC knockout pigs	TALEN	Infertility and colon-cancer models	[78]
Multiple mouse gene modifications	CRISPR	Immunodeficiency models	[79]
[Ppar-γ + Rag1] knockout monkeys	CRISPR	Ppar-γ + Rag1 disease associations	[80]

CRISPR: Clustered regulatory interspaced short palindromic repeat; ICF: Immunodeficiency, centromeric region instability, facial anomalies syndrome; iPS: Induced pluripotent stem cells; PCS (MVA): Premature chromatid separation with mosaic variegated aneuploidy; PGE: Precision genome editing; TALEN: Transcription activator-like effector nuclease; XSCID: X-linked severe combined immunodeficiency; ZFN: Zinc-finger nuclease.

5. Conclusion

Research is characterized by change and unpredictability, both generally and with regard to the technology marketplace. HDX-MS, ABPP and targetable, nuclease-mediated PGE exemplify progressive technologies whose current reach and information yield position them as increasingly important and useful architects of the evolving discovery landscape and portend even more significant contributions to lead identification/characterization in drug discovery.

6. Expert opinion

Insights already afforded by the relatively new technologies of HDX-MS, ABPP and targetable nuclease-mediated PGE do not allow their dismissal by the discovery community as mere academic ‘solutions looking for problems’ [81]. Rather, the author opines that these evolving technologies will continue to strengthen and diversify drug discovery links with the disciplines of structural biology, chemical proteomics and functional genomics to enhance our understanding of what potential therapeutics do and how they act in more disease-relevant experimental contexts prior to the huge investment ramp-up mandated by human studies and so often lost along the clinical development pathway. HDX-MS, ABPP and targetable nuclease-mediated PGE can be applied to interrogate biological pathways and information systems (e.g., proteomes, disease and drug-sensitive networks, functional genome-wide screening) in multiplex fashion, since they span across levels of biological organization (from molecules to mammals). This feature opens avenues beyond the reductionist strictures of target-focused pharmacophoric/chemotype searching, an ingrained practice often driven more by feasibility than by evidence of its being a route to competitive advantage in drug discovery. This proposition is enticing, given that integrative, systems-based phenotypic screening was the source of most first-in-class small-molecule medications approved between 1999 and 2008 [1]. The fact that insufficient efficacy was the most frequent cause of Phase II clinical failures occurring during 2008 – 2011 and regulatory rejection of new molecular entities by the United States FDA during 2000 – 2012 [82,83] strongly suggests that limitations of established disease models within canonical, target-based discovery paradigms have jeopardized clinical success rates by providing candidate identification/characterization insufficient for clinical development [84]. By extension, enhanced understanding of therapeutically relevant biology and pharmacology of lead compounds as afforded by HDX-MS and ABPP in the preclinical context of pathologically relevant, PGE-derived disease models should help inform critical-path ‘go/no-go’ decision-making around compound attributes (e.g., efficacy, selectivity, mode of action, off-target interactions) important to populating discovery pipelines, assessing the potential for lead advancement into development, and reducing the inherently high risk associated with drug innovation.

The author posits that continued advances along these and other lines will intensify ongoing efforts to broaden the range of applications for HDX-MS, ABPP and nuclease-mediated PGE and will enhance the benefits these technologies bring to drug discovery. New applications should be fostered by relationships to both biological value and therapeutic impact – not by hype or technophilia – and would likely encompass the following:

6.1 Hydrogen–deuterium exchange mass spectrometry

Biopharmaceutical analysis and quality control; identification of new, functionally important protein ligand-binding (e.g., allosteric) sites to guide structure-based drug design; definition of protein structural pathology (e.g., aggregation domains) suggesting new therapeutic modalities.

6.2 Activity-based protein profiling

Biomarker discovery for in vivo diagnostics/imaging; annotation of the (therapeutic) activity of natural products and deconvolution of their biological targets; screening drugs for promiscuity/off-target effects; disease-related proteasome mapping in living systems.

6.3 Nuclease-mediated PGE

Defining the genetic etiology of complex disease phenotypes in vivo; generating tissue-specific, knock-in ‘humanized’ disease models; translating emerging data suggestive of gene-based correction therapies (e.g., mutation-focused drugs) [85,86]; producing mammalian cells, tissues and organs for regenerative-medicine and synthetic-biology applications; high-resolution mapping of disease interactomes.

Even with their rich information yields and future potential, HDX-MS, ABPP and/or nuclease-mediated PGE per se cannot be expected to solve current pharma-industry woes [2]. Nor are these technologies necessarily relevant or applicable to every discovery program, for inappropriate utilization of even the most time-honored techniques can hinder research progress. Every technology has limits and can be improved/refined. In this latter regard, recent HDX-MS procedural and automation refinements have reduced analysis time, facilitated data acquisition/handling, enhanced spatial and temporal resolution, and allowed examination of heterogeneous protein samples, with ongoing initiatives to reduce the dependence on available protein structural information for interpretation of peptide-level data [15-19]. While definition of critical amino acid residues and binding poises essential to ligand engagement and therapeutic functional output by protein drug targets has proven useful to guide lead evolution and therapeutics design [87], state-of-the-art HDX-MS is at the amino acid level and does not attain the atomic resolution of X-ray crystallography. Advancements in probe design and diversity and detection instrumentation have improved ABPP sensitivity for low-abundance proteins and made it possible to envision in vitro and in vivo ABPP applications to protein classes (ion channels, receptors, etc.) beyond those enzymes now amenable to analysis [37,38,41]. In its current elaboration, ABPP perhaps best serves as a proteomic-level extension of – rather than a substitute for – direct enzyme assay to determine inhibitor constants for detailed structure–activity profiling. Improvements in monitoring genome-editing quality control (i.e., PGE nuclease specificity and target-site recognition) and the capacity and accuracy of high-throughput sequencing have enhanced the robustness, uniformity and reproducibility of nuclease-mediated PGE, as have emerging details about the structural basis of CRISPR technology [88-90]. Although CRISPR-Cas9 is now considered the leading PGE platform because of its accessibility, applications versatility and simplicity as compared to the older ZFN and TALEN PGE technologies that require complicated protein engineering to ensure correct zinc-finger targeting, CRISPR-Cas9 faces potential hurdles (especially for direct therapeutic application) from off-target guide-RNA binding to DNA and the nature of Cas9 as a bacterial protein that could provoke an immune response in mammals [91]. Complementarities among HDX-MS, ABPP, nuclease-mediated PGE and other technologies in structural biology, chemoproteomics, and functional genomics and limitations associated with each technology have invited concerted application of experimental approaches to amplify information yield relevant to drug discovery. For example, integration of HDX-MS with several other structural biology and protein-analysis techniques (mass, infrared and nuclear magnetic resonance spectrometries; circular dichroism; calorimetry; X-ray crystallography; peptide mapping) was essential for demonstrating that a candidate biosimilar was compatible with an originator monoclonal-antibody drug, a key requirement for regulatory approval [92].

As John Donne (1572 – 1631) reminds us: ‘No man is an island.’ Now more than ever, drug discovery is a group endeavor requiring both diverse constituencies that share professional and technical knowledge and appreciable front-loaded financial and intellectual resources. In this regard, it is noteworthy that HDX-MS, ABPP, nuclease-mediated PGE originated from the public (academic/research institute) sector [93-95], which would suggest opportunities for precompetitive public–private (i.e., industrial) collaboration to optimize these technologies and expand their burgeoning drug-discovery applications. Despite current opportunities such as the United States National Institute of Health’s Shared Instrumentation Grant [96] and Instrument Development for Biomedical Applications Programs [97] and non-governmental initiatives such as the Biomedical Resource and Technology Development grants offered in the United Kingdom by the Wellcome Trust [98], support for fundamental research on discovery-relevant technology has been historically limited. Increased levels of funding and greater numbers of scientists dedicated to actualizing and refining these (and other developing) discovery technologies are urgently needed and could encourage research collaborations and facilitate wider technology access by researchers [2]. Precompetitive collaborations with (big) pharma and other companies having vested interest in methods application and/or commercialization could serve as another spur to technology actualization for drug discovery. Startup contract organizations for nuclease-mediated PGE are but one current embodiment of this thinking [99].

The only real success in drug discovery is delivery of a therapeutic into the clinic as an approved medication that addresses an unmet medical need safely and efficaciously. Integration of any technology into a discovery effort is no guarantee that this litmus test will be met. Nonetheless, as illustrated herein, application of technological advancements early in the discovery process can provide novel information that enriches candidate profiling in more disease-relevant models and deepens our understanding of disease processes. The beleaguered state of the pharmaceutical industry clearly requires discovery approaches beyond legacy groupthink that are open to developing, supporting and embracing innovative technologies. To quote the Canadian philosopher Marshall McLuhan (1911 – 1980): ‘It is the framework which changes with each new technology and not just the picture within the frame.’

Declaration of interest

D Janero has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Acknowledgment

The expert bibliographic assistance of Mr. Mitchel Ayer is gratefully acknowledged.

Notes