Crystallographic fragment screening in academic cancer drug discovery.

Fragment-based drug discovery (FBDD) has brought several drugs to the clinic, notably to target proteins once considered to be challenging, or even undruggable. Screening in FBDD relies upon observing and/or measuring weak (millimolar-scale) binding events using biophysical techniques or crystallographic fragment screening. This latter structural approach provides no information about binding affinity but can reveal binding mode and atomic detail on protein-fragment interactions to accelerate hit-to-lead development. In recent years, high-throughput platforms have been developed at synchrotron facilities to screen thousands of fragment-soaked crystals. However, using accessible manual techniques it is possible to run informative, smaller-scale screens within an academic lab setting. This chapter describes general protocols for home laboratory-scale fragment screening, from fragment soaking through to structure solution and, where appropriate, signposts to background, protocols or alternatives elsewhere.

Fragment expansion with NUDELs - poised DNA-encoded libraries.

Optimisation of the affinity of lead compounds is a critical challenge in the identification of drug candidates and chemical probes and is a process that takes many years. Fragment-based drug discovery has become established as one of the methods of choice for drug discovery starting with small, low affinity compounds. Due to their low affinity, the evolution of fragments to desirable levels of affinity is often a key challenge. The accepted best method for increasing the potency of fragments is by iterative fragment growing, which can be very time consuming and complex. Here, we introduce a paradigm for fragment hit optimisation using poised DNA-encoded chemical libraries (DELs). The synthesis of a poised DEL, a partially constructed library that retains a reactive handle, allows the coupling of any active fragment for a specific target protein, allowing rapid discovery of potent ligands. This is illustrated for bromodomain-containing protein 4 (BRD4), in which a weakly binding fragment was coupled to a 42-member poised DEL via Suzuki-Miyaura cross coupling resulting in the identification of an inhibitor with 51 nM affinity in a single step, representing an increase in potency of several orders of magnitude from an original fragment. The potency of the compound was shown to arise from the synergistic combination of substructures, which would have been very difficult to discover by any other method and was rationalised by X-ray crystallography. The compound showed attractive lead-like properties suitable for further optimisation and demonstrated BRD4-dependent cellular pharmacology. This work demonstrates the power of poised DELs to rapidly optimise fragments, representing an attractive generic approach to drug discovery.

Targeting Cytotoxic Agents through EGFR-Mediated Covalent Binding and Release.

A major drawback of cytotoxic chemotherapy is the lack of selectivity toward noncancerous cells. The targeted delivery of cytotoxic drugs to tumor cells is a longstanding goal in cancer research. We proposed that covalent inhibitors could be adapted to deliver cytotoxic agents, conjugated to the ß-position of the Michael acceptor, via an addition-elimination mechanism promoted by covalent binding. Studies on model systems showed that conjugated 5-fluorouracil (5FU) could be released upon thiol addition in relevant time scales. A series of covalent epidermal growth factor receptor (EGFR) inhibitors were synthesized as their 5FU derivatives. Achieving the desired release of 5FU was demonstrated to depend on the electronics and geometry of the compounds. Mass spectrometry and NMR studies demonstrated an anilinoquinazoline acrylate ester conjugate bound to EGFR with the release of 5FU. This work establishes that acrylates can be used to release conjugated molecules upon covalent binding to proteins and could be used to develop targeted therapeutics.

Cryo-EM structure of SKP1-SKP2-CKS1 in complex with CDK2-cyclin A-p27KIP1.

p27KIP1 (cyclin-dependent kinase inhibitor 1B, p27) is a member of the CIP/KIP family of CDK (cyclin dependent kinase) regulators that inhibit cell cycle CDKs. p27 phosphorylation by CDK1/2, signals its recruitment to the SCFSKP2 (S-phase kinase associated protein 1 (SKP1)-cullin-SKP2) E3 ubiquitin ligase complex for proteasomal degradation. The nature of p27 binding to SKP2 and CKS1 was revealed by the SKP1-SKP2-CKS1-p27 phosphopeptide crystal structure. Subsequently, a model for the hexameric CDK2-cyclin A-CKS1-p27-SKP1-SKP2 complex was proposed by overlaying an independently determined CDK2-cyclin A-p27 structure. Here we describe the experimentally determined structure of the isolated CDK2-cyclin A-CKS1-p27-SKP1-SKP2 complex at 3.4 Å global resolution using cryogenic electron microscopy. This structure supports previous analysis in which p27 was found to be structurally dynamic, transitioning from disordered to nascent secondary structure on target binding. We employed 3D variability analysis to further explore the conformational space of the hexameric complex and uncovered a previously unidentified hinge motion centred on CKS1. This flexibility gives rise to open and closed conformations of the hexameric complex that we propose may contribute to p27 regulation by facilitating recognition with SCFSKP2. This 3D variability analysis further informed particle subtraction and local refinement approaches to enhance the local resolution of the complex.

Exiting the tunnel of uncertainty: crystal soak to validated hit.

Crystallographic fragment screens provide an efficient and effective way to identify small-molecule ligands of a crystallized protein. Due to their low molecular weight, such hits tend to have low, often unquantifiable, affinity for their target, complicating the twin challenges of validating the hits as authentic solution-phase ligands of the target and identifying the `best' hit(s) for further elaboration. In this article, approaches that address these challenges are assessed. Using retrospective analysis of a recent ATAD2 hit-identification campaign, alongside other examples of successful fragment-screening campaigns, it is suggested that hit validation and prioritization are best achieved by a `triangulation' approach in which the results of multiple available biochemical and biophysical techniques are correlated to develop qualitative structure-activity relationships (SARs). Such qualitative SARs may indeed be the only means by which to navigate a project through the tunnel of uncertainty that prevails before on-scale biophysical, biochemical and/or biological measurements become possible.

Mapping Ligand Interactions of Bromodomains BRD4 and ATAD2 with FragLites and PepLites─Halogenated Probes of Druglike and Peptide-like Molecular Interactions.

The development of ligands for biological targets is critically dependent on the identification of sites on proteins that bind molecules with high affinity. A set of compounds, called FragLites, can identify such sites, along with the interactions required to gain affinity, by X-ray crystallography. We demonstrate the utility of FragLites in mapping the binding sites of bromodomain proteins BRD4 and ATAD2 and demonstrate that FragLite mapping is comparable to a full fragment screen in identifying ligand binding sites and key interactions. We extend the FragLite set with analogous compounds derived from amino acids (termed PepLites) that mimic the interactions of peptides. The output of the FragLite maps is shown to enable the development of ligands with leadlike potency. This work establishes the use of FragLite and PepLite screening at an early stage in ligand discovery allowing the rapid assessment of tractability of protein targets and informing downstream hit-finding.

Build-Couple-Transform: A Paradigm for Lead-like Library Synthesis with Scaffold Diversity.

High-throughput screening provides one of the most common ways of finding hit compounds. Lead-like libraries, in particular, provide hits with compatible functional groups and vectors for structural elaboration and physical properties suitable for optimization. Library synthesis approaches can lead to a lack of chemical diversity because they employ parallel derivatization of common building blocks using single reaction types. We address this problem through a "build-couple-transform" paradigm for the generation of lead-like libraries with scaffold diversity. Nineteen transformations of a 4-oxo-2-butenamide scaffold template were optimized, including 1,4-cyclizations, 3,4-cyclizations, reductions, and 1,4-additions. A pool-transformation approach efficiently explored the scope of these transformations for nine different building blocks and synthesized a >170-member library with enhanced chemical space coverage and favorable drug-like properties. Screening revealed hits against CDK2. This work establishes the build-couple-transform concept for the synthesis of lead-like libraries and provides a differentiated approach to libraries with significantly enhanced scaffold diversity.

Parallel Optimization of Potency and Pharmacokinetics Leading to the Discovery of a Pyrrole Carboxamide ERK5 Kinase Domain Inhibitor.

The nonclassical extracellular signal-related kinase 5 (ERK5) mitogen-activated protein kinase pathway has been implicated in increased cellular proliferation, migration, survival, and angiogenesis; hence, ERK5 inhibition may be an attractive approach for cancer treatment. However, the development of selective ERK5 inhibitors has been challenging. Previously, we described the development of a pyrrole carboxamide high-throughput screening hit into a selective, submicromolar inhibitor of ERK5 kinase activity. Improvement in the ERK5 potency was necessary for the identification of a tool ERK5 inhibitor for target validation studies. Herein, we describe the optimization of this series to identify nanomolar pyrrole carboxamide inhibitors of ERK5 incorporating a basic center, which suffered from poor oral bioavailability. Parallel optimization of potency and in vitro pharmacokinetic parameters led to the identification of a nonbasic pyrazole analogue with an optimal balance of ERK5 inhibition and oral exposure.

An Alkynylpyrimidine-Based Covalent Inhibitor That Targets a Unique Cysteine in NF-κB-Inducing Kinase.

NF-κB-inducing kinase (NIK) is a key enzyme in the noncanonical NF-κB pathway, of interest in the treatment of a variety of diseases including cancer. Validation of NIK as a drug target requires potent and selective inhibitors. The protein contains a cysteine residue at position 444 in the back pocket of the active site, unique within the kinome. Analysis of existing inhibitor scaffolds and early structure-activity relationships (SARs) led to the design of C444-targeting covalent inhibitors based on alkynyl heterocycle warheads. Mass spectrometry provided proof of the covalent mechanism, and the SAR was rationalized by computational modeling. Profiling of more potent analogues in tumor cell lines with constitutively activated NIK signaling induced a weak antiproliferative effect, suggesting that kinase inhibition may have limited impact on cancer cell growth. This study shows that alkynyl heterocycles are potential cysteine traps, which may be employed where common Michael acceptors, such as acrylamides, are not tolerated.

Structure-Based Design of Potent and Orally Active Isoindolinone Inhibitors of MDM2-p53 Protein-Protein Interaction.

Inhibition of murine double minute 2 (MDM2)-p53 protein-protein interaction with small molecules has been shown to reactivate p53 and inhibit tumor growth. Here, we describe rational, structure-guided, design of novel isoindolinone-based MDM2 inhibitors. MDM2 X-ray crystallography, quantum mechanics ligand-based design, and metabolite identification all contributed toward the discovery of potent in vitro and in vivo inhibitors of the MDM2-p53 interaction with representative compounds inducing cytostasis in an SJSA-1 osteosarcoma xenograft model following once-daily oral administration.

Discriminative SKP2 Interactions with CDK-Cyclin Complexes Support a Cyclin A-Specific Role in p27KIP1 Degradation.

The SCFSKP2 ubiquitin ligase relieves G1 checkpoint control of CDK-cyclin complexes by promoting p27KIP1 degradation. We describe reconstitution of stable complexes containing SKP1-SKP2 and CDK1-cyclin B or CDK2-cyclin A/E, mediated by the CDK regulatory subunit CKS1. We further show that a direct interaction between a SKP2 N-terminal motif and cyclin A can stabilize SKP1-SKP2-CDK2-cyclin A complexes in the absence of CKS1. We identify the SKP2 binding site on cyclin A and demonstrate the site is not present in cyclin B or cyclin E. This site is distinct from but overlapping with features that mediate binding of p27KIP1 and other G1 cyclin regulators to cyclin A. We propose that the capacity of SKP2 to engage with CDK2-cyclin A by more than one structural mechanism provides a way to fine tune the degradation of p27KIP1 and distinguishes cyclin A from other G1 cyclins to ensure orderly cell cycle progression.

Crystallographic and electrophilic fragment screening of the SARS-CoV-2 main protease.

COVID-19, caused by SARS-CoV-2, lacks effective therapeutics. Additionally, no antiviral drugs or vaccines were developed against the closely related coronavirus, SARS-CoV-1 or MERS-CoV, despite previous zoonotic outbreaks. To identify starting points for such therapeutics, we performed a large-scale screen of electrophile and non-covalent fragments through a combined mass spectrometry and X-ray approach against the SARS-CoV-2 main protease, one of two cysteine viral proteases essential for viral replication. Our crystallographic screen identified 71 hits that span the entire active site, as well as 3 hits at the dimer interface. These structures reveal routes to rapidly develop more potent inhibitors through merging of covalent and non-covalent fragment hits; one series of low-reactivity, tractable covalent fragments were progressed to discover improved binders. These combined hits offer unprecedented structural and reactivity information for on-going structure-based drug design against SARS-CoV-2 main protease.

Paradoxical activation of the protein kinase-transcription factor ERK5 by ERK5 kinase inhibitors.

The dual protein kinase-transcription factor, ERK5, is an emerging drug target in cancer and inflammation, and small-molecule ERK5 kinase inhibitors have been developed. However, selective ERK5 kinase inhibitors fail to recapitulate ERK5 genetic ablation phenotypes, suggesting kinase-independent functions for ERK5. Here we show that ERK5 kinase inhibitors cause paradoxical activation of ERK5 transcriptional activity mediated through its unique C-terminal transcriptional activation domain (TAD). Using the ERK5 kinase inhibitor, Compound 26 (ERK5-IN-1), as a paradigm, we have developed kinase-active, drug-resistant mutants of ERK5. With these mutants, we show that induction of ERK5 transcriptional activity requires direct binding of the inhibitor to the kinase domain. This in turn promotes conformational changes in the kinase domain that result in nuclear translocation of ERK5 and stimulation of gene transcription. This shows that both the ERK5 kinase and TAD must be considered when assessing the role of ERK5 and the effectiveness of anti-ERK5 therapeutics.

Identification of a novel orally bioavailable ERK5 inhibitor with selectivity over p38α and BRD4.

Extracellular regulated kinase 5 (ERK5) signalling has been implicated in driving a number of cellular phenotypes including endothelial cell angiogenesis and tumour cell motility. Novel ERK5 inhibitors were identified using high throughput screening, with a series of pyrrole-2-carboxamides substituted at the 4-position with an aroyl group being found to exhibit IC50 values in the micromolar range, but having no selectivity against p38α MAP kinase. Truncation of the N-substituent marginally enhanced potency (∼3-fold) against ERK5, but importantly attenuated inhibition of p38α. Systematic variation of the substituents on the aroyl group led to the selective inhibitor 4-(2-bromo-6-fluorobenzoyl)-N-(pyridin-3-yl)-1H-pyrrole-2-carboxamide (IC50 0.82 µM for ERK5; IC50 > 120 µM for p38α). The crystal structure (PDB 5O7I) of this compound in complex with ERK5 has been solved. This compound was orally bioavailable and inhibited bFGF-driven Matrigel plug angiogenesis and tumour xenograft growth. The selective ERK5 inhibitor described herein provides a lead for further development into a tool compound for more extensive studies seeking to examine the role of ERK5 signalling in cancer and other diseases.

FragLites-Minimal, Halogenated Fragments Displaying Pharmacophore Doublets. An Efficient Approach to Druggability Assessment and Hit Generation.

Identifying ligand binding sites on proteins is a critical step in target-based drug discovery. Current approaches to this require resource-intensive screening of large libraries of lead-like or fragment molecules. Here, we describe an efficient and effective experimental approach to mapping interaction sites using a set of halogenated compounds expressing paired hydrogen-bonding motifs, termed FragLites. The FragLites identify productive drug-like interactions, which are identified sensitively and unambiguously by X-ray crystallography, exploiting the anomalous scattering of the halogen substituent. This mapping of protein interaction surfaces provides an assessment of druggability and can identify efficient start points for the de novo design of hit molecules incorporating the interacting motifs. The approach is illustrated by mapping cyclin-dependent kinase 2, which successfully identifies orthosteric and allosteric sites. The hits were rapidly elaborated to develop efficient lead-like molecules. Hence, the approach provides a new method of identifying ligand sites, assessing tractability and discovering new leads.

Differences in the Conformational Energy Landscape of CDK1 and CDK2 Suggest a Mechanism for Achieving Selective CDK Inhibition.

Dysregulation of the cell cycle characterizes many cancer subtypes, providing a rationale for developing cyclin-dependent kinase (CDK) inhibitors. Potent CDK2 inhibitors might target certain cancers in which CCNE1 is amplified. However, current CDK2 inhibitors also inhibit CDK1, generating a toxicity liability. We have used biophysical measurements and X-ray crystallography to investigate the ATP-competitive inhibitor binding properties of cyclin-free and cyclin-bound CDK1 and CDK2. We show that these kinases can readily be distinguished by such inhibitors when cyclin-free, but not when cyclin-bound. The basis for this discrimination is unclear from either inspection or molecular dynamics simulation of ligand-bound CDKs, but is reflected in the contacts made between the kinase N- and C-lobes. We conclude that there is a subtle but profound difference between the conformational energy landscapes of cyclin-free CDK1 and CDK2. The unusual properties of CDK1 might be exploited to differentiate CDK1 from other CDKs in future cancer therapeutic design.

Tuning the Binding Affinity and Selectivity of Perfluoroaryl-Stapled Peptides by Cysteine-Editing.

A growing number of approaches to "staple" α-helical peptides into a bioactive conformation using cysteine cross-linking are emerging. Here, the replacement of l-cysteine with "cysteine analogues" in combinations of different stereochemistry, side chain length and beta-carbon substitution, is explored to examine the influence that the thiol-containing residue(s) has on target protein binding affinity in a well-explored model system, p53-MDM2/MDMX, which is constituted by the interaction of the tumour suppressor protein p53 and proteins MDM2 and MDMX, which regulate p53 activity. In some cases, replacement of one or more l-cysteine residues afforded significant changes in the measured binding affinity and target selectivity of the peptide. Computationally constructed homology models indicate that some modifications, such as incorporating two d-cysteine residues, favourably alter the positions of key functional amino acid side chains, which is likely to cause changes in binding affinity, in agreement with measured surface plasmon resonance data.

Validating and enabling phosphoglycerate dehydrogenase (PHGDH) as a target for fragment-based drug discovery in PHGDH-amplified breast cancer.

3-Phosphoglycerate dehydrogenase (PHGDH) has recently been identified as an attractive target in cancer therapy as it links upregulated glycolytic flux to increased biomass production in cancer cells. PHGDH catalyses the first step in the serine synthesis pathway and thus diverts glycolytic flux into serine synthesis. We have used siRNA-mediated suppression of PHGDH expression to show that PHGDH is a potential therapeutic target in PHGDH-amplified breast cancer. Knockdown caused reduced proliferation in the PHGDH-amplified cell line MDA-MB-468, whereas breast cancer cells with low PHGDH expression or with elevated PHGDH expression in the absence of genomic amplification were not affected. As a first step towards design of a chemical probe for PHGDH, we report a fragment-based drug discovery approach for the identification of PHGDH inhibitors. We designed a truncated PHGDH construct that gave crystals which diffracted to high resolution, and could be used for fragment soaking. 15 fragments stabilising PHGDH were identified using a thermal shift assay and validated by X-ray crystallography and ITC competition experiments to exhibit 1.5-26.2 mM affinity for PHGDH. A structure-guided fragment growing approach was applied to the PHGDH binders from the initial screen, yielding greater understanding of the binding site and suggesting routes to achieve higher affinity NAD-competitive inhibitors.

Identification of a novel ligand for the ATAD2 bromodomain with selectivity over BRD4 through a fragment growing approach.

ATAD2 is an ATPase that is overexpressed in a variety of cancers and associated with a poor patient prognosis. This protein has been suggested to function as a cofactor for a range of transcription factors, including the proto-oncogene MYC and the androgen receptor. ATAD2 comprises an ATPase domain, implicated in chromatin remodelling, and a bromodomain which allows it to interact with acetylated histone tails. Dissection of the functional roles of these two domains would benefit from the availability of selective, cell-permeable pharmacological probes. An in silico evaluation of the 3D structures of various bromodomains suggested that developing small molecule ligands for the bromodomain of ATAD2 is likely to be challenging, although recent reports have shown that ATAD2 bromodomain ligands can be identified. We report a structure-guided fragment-based approach to identify lead compounds for ATAD2 bromodomain inhibitor development. Our findings indicate that the ATAD2 bromodomain can accommodate fragment hits (Mr < 200) that yield productive structure-activity relationships, and structure-guided design enabled the introduction of selectivity over BRD4.

Differential Regulation of G1 CDK Complexes by the Hsp90-Cdc37 Chaperone System.

Selective recruitment of protein kinases to the Hsp90 system is mediated by the adaptor co-chaperone Cdc37. We show that assembly of CDK4 and CDK6 into protein complexes is differentially regulated by the Cdc37-Hsp90 system. Like other Hsp90 kinase clients, binding of CDK4/6 to Cdc37 is blocked by ATP-competitive inhibitors. Cdc37-Hsp90 relinquishes CDK6 to D3- and virus-type cyclins and to INK family CDK inhibitors, whereas CDK4 is relinquished to INKs but less readily to cyclins. p21CIP1 and p27KIP1 CDK inhibitors are less potent than the INKs at displacing CDK4 and CDK6 from Cdc37. However, they cooperate with the D-type cyclins to generate CDK4/6-containing ternary complexes that are resistant to cyclin D displacement by Cdc37, suggesting a molecular mechanism to explain the assembly factor activity ascribed to CIP/KIP family members. Overall, our data reveal multiple mechanisms whereby the Hsp90 system may control formation of CDK4- and CDK6-cyclin complexes under different cellular conditions.