Frank M. Boeckler

Principles and Applications of Halogen Bonding in Medicinal Chemistry and Chemical Biology

Halogen bonding has been known in material science for decades, but until recently, halogen bonds in protein-ligand interactions were largely the result of serendipitous discovery rather than rational design. In this Perspective, we provide insights into the phenomenon of halogen bonding, with special focus on its role in drug discovery. We summarize the theoretical background defining its strength and directionality, provide a systematic analysis of its occurrence and interaction geometries in protein-ligand complexes, and give recent examples where halogen bonding has been successfully harnessed for lead identification and optimization. In light of these data, we discuss the potential and limitations of exploiting halogen bonds for molecular recognition and rational drug design.

Targeted rescue of a destabilized mutant of p53 by an in silico screened drug

The tumor suppressor p53 is mutationally inactivated in ≈50% of human cancers. Approximately one-third of the mutations lower the melting temperature of the protein, leading to its rapid denaturation. Small molecules that bind to those mutants and stabilize them could be effective anticancer drugs. The mutation Y220C, which occurs in ≈75,000 new cancer cases per annum, creates a surface cavity that destabilizes the protein by 4 kcal/mol, at a site that is not functional. We have designed a series of binding molecules from an in silico analysis of the crystal structure using virtual screening and rational drug design. One of them, a carbazole derivative (PhiKan083), binds to the cavity with a dissociation constant of ≈150 μM. It raises the melting temperature of the mutant and slows down its rate of denaturation. We have solved the crystal structure of the protein–PhiKan083 complex at 1.5-Å resolution. The structure implicates key interactions between the protein and ligand and conformational changes that occur on binding, which will provide a basis for lead optimization. The Y220C mutant is an excellent “druggable” target for developing and testing novel anticancer drugs based on protein stabilization. We point out some general principles in relationships between binding constants, raising of melting temperatures, and increase of protein half-lives by stabilizing ligands.

Halogen-Enriched Fragment Libraries as Leads for Drug Rescue of Mutant P53.

The destabilizing p53 cancer mutation Y220C creates a druggable surface crevice. We developed a strategy exploiting halogen bonding for lead discovery to stabilize the mutant with small molecules. We designed halogen-enriched fragment libraries (HEFLibs) as starting points to complement classical approaches. From screening of HEFLibs and subsequent structure-guided design, we developed substituted 2-(aminomethyl)-4-ethynyl-6-iodophenols as p53-Y220C stabilizers. Crystal structures of their complexes highlight two key features: (i) a central scaffold with a robust binding mode anchored by halogen bonding of an iodine with a main-chain carbonyl and (ii) an acetylene linker, enabling the targeting of an additional subsite in the crevice. The best binders showed induction of apoptosis in a human cancer cell line with homozygous Y220C mutation. Our structural and biophysical data suggest a more widespread applicability of HEFLibs in drug discovery.

Using halogen bonds to address the protein backbone: a systematic evaluation

Halogen bonds are specific embodiments of the sigma hole bonding paradigm. They represent directional interactions between the halogens chlorine, bromine, or iodine and an electron donor as binding partner. Using quantum chemical calculations at the MP2 level, we systematically explore how they can be used in molecular design to address the omnipresent carbonyls of the protein backbone. We characterize energetics and directionality and elucidate their spatial variability in sub-optimal geometries that are expected to occur in protein–ligand complexes featuring a multitude of concomitant interactions. By deriving simple rules, we aid medicinal chemists and chemical biologists in easily exploiting them for scaffold decoration and design. Our work shows that carbonyl–halogen bonds may be used to expand the patentable medicinal chemistry space, redefining halogens as key features. Furthermore, this data will be useful for implementing halogen bonds into pharmacophore models or scoring functions making the QM information available for automatic molecular recognition in virtual high throughput screening.

Kinetic mechanism of p53 oncogenic mutant aggregation and its inhibition

Aggregation of destabilized mutants of the tumor suppressor p53 is a major route for its loss of activity. In order to assay drugs that inhibit aggregation of p53, we established the basic kinetics of aggregation of its core domain, using the mutant Y220C that has a mutation-induced, druggable cavity. Aggregation monitored by light scattering followed lag kinetics. Electron microscopy revealed the formation of small aggregates that subsequently grew to larger amorphous aggregates. The kinetics of aggregation produced surprising results: progress curves followed either by the binding of Thioflavin T or the fluorescence of the protein at 340 nm fitted well to simple two-step sequential first-order lag kinetics with rate constants k1 and k2 that were independent of protein concentration, and not to classical nucleation-growth. We suggest a mechanism of first-order formation of an aggregation competent state as being rate determining followed by rapid polymerization with the higher order kinetics. By measuring the inhibition kinetics of k1 and k2, we resolved that the process with the higher rate constant followed that of the lower. Further, there was only partial inhibition of k1 and k2, which showed two parallel pathways of aggregation, one via a state that requires unfolding of the protein and the other of partial unfolding with the ligand still bound. Inhibition kinetics of ligands provides a useful tool for probing an aggregation mechanism.

Tri- and Tetrasubstituted Pyrazole Derivates: Regioisomerism Switches Activity from p38MAP Kinase to Important Cancer Kinases

In the course of searching for new p38α MAP kinase inhibitors, we found that the regioisomeric switch from 3-(4-fluorophenyl)-4-(pyridin-4-yl)-1-(aryl)-1H-pyrazol-5-amine to 4-(4-fluorophenyl)-3-(pyridin-4-yl)-1-(aryl)-1H-pyrazol-5-amine led to an almost complete loss of p38α inhibition, but they showed activity against important cancer kinases. Among the tested derivatives, 4-(4-fluorophenyl)-3-(pyridin-4-yl)-1-(2,4,6-trichlorophenyl)-1H-pyrazol-5-amine (6a) exhibited the best activity, with IC(50) in the nanomolar range against Src, B-Raf wt, B-Raf V600E, EGFRs, and VEGFR-2, making it a good lead for novel anticancer programs.

Evaluation and Optimization of Virtual Screening Workflows with DEKOIS 2.0-A Public Library of Challenging Docking Benchmark Sets

The application of molecular benchmarking sets helps to assess the actual performance of virtual screening (VS) workflows. To improve the efficiency of structure-based VS approaches, the selection and optimization of various parameters can be guided by benchmarking. With the DEKOIS 2.0 library, we aim to further extend and complement the collection of publicly available decoy sets. Based on BindingDB bioactivity data, we provide 81 new and structurally diverse benchmark sets for a wide variety of different target classes. To ensure a meaningful selection of ligands, we address several issues that can be found in bioactivity data. We have improved our previously introduced DEKOIS methodology with enhanced physicochemical matching, now including the consideration of molecular charges, as well as a more sophisticated elimination of latent actives in the decoy set (LADS). We evaluate the docking performance of Glide, GOLD, and AutoDock Vina with our data sets and highlight existing challenges for VS tools. All DEKOIS 2.0 benchmark sets will be made accessible at http://www.dekois.com.

Lithocholic acid is an endogenous inhibitor of MDM4 and MDM2

The proteins MDM2 and MDM4 are key negative regulators of the tumor suppressor protein p53, which are frequently upregulated in cancer cells. They inhibit the transactivation activity of p53 by binding separately or in concert to its transactivation domain. MDM2 is also a ubiquitin ligase that leads to the degradation of p53. Accordingly, MDM2 and MDM4 are important targets for drugs to inhibit their binding to p53. We found from in silico screening and confirmed by experiment that lithocholic acid (LCA) binds to the p53 binding sites of both MDM2 and MDM4 with a fivefold preference for MDM4. LCA is an endogenous steroidal bile acid, variously reported to have both carcinogenic and apoptotic activities. The comparison of LCA effects on apoptosis in HCT116 p53+/+ vs. p53-/- cells shows a predominantly p53-mediated induction of caspase-3/7. The dissociation constants are in the μM region, but only modest inhibition of binding of MDM2 and MDM4 is required to negate their upregulation because they have to compete with transcriptional coactivator p300 for binding to p53. Binding was weakened by structural changes in LCA, and so it may be a natural ligand of MDM2 and MDM4, raising the possibility that MDM proteins may be sensors for specific steroids.

DEKOIS: demanding evaluation kits for objective in silico screening--a versatile tool for benchmarking docking programs and scoring functions.

For widely applied in silico screening techniques success depends on the rational selection of an appropriate method. We herein present a fast, versatile, and robust method to construct demanding evaluation kits for objective in silico screening (DEKOIS). This automated process enables creating tailor-made decoy sets for any given sets of bioactives. It facilitates a target-dependent validation of docking algorithms and scoring functions helping to save time and resources. We have developed metrics for assessing and improving decoy set quality and employ them to investigate how decoy embedding affects docking. We demonstrate that screening performance is target-dependent and can be impaired by latent actives in the decoy set (LADS) or enhanced by poor decoy embedding. The presented method allows extending and complementing the collection of publicly available high quality decoy sets toward new target space. All present and future DEKOIS data sets will be made accessible at www.dekois.com.

Targeting the Gatekeeper MET146 of C-Jun N-Terminal Kinase 3 Induces a Bivalent Halogen/Chalcogen Bond

We target the gatekeeper MET146 of c-Jun N-terminal kinase 3 (JNK3) to exemplify the applicability of X···S halogen bonds in molecular design using computational, synthetic, structural and biophysical techniques. In a designed series of aminopyrimidine-based inhibitors, we unexpectedly encounter a plateau of affinity. Compared to their QM-calculated interaction energies, particularly bromine and iodine fail to reach the full potential according to the size of their σ-hole. Instead, mutation of the gatekeeper residue into leucine, alanine, or threonine reveals that the heavier halides can significantly influence selectivity in the human kinome. Thus, we demonstrate that, although the choice of halogen may not always increase affinity, it can still be relevant for inducing selectivity. Determining the crystal structure of the iodine derivative in complex with JNK3 (4X21) reveals an unusual bivalent halogen/chalcogen bond donated by the ligand and the back-pocket residue MET115. Incipient repulsion from the too short halogen bond increases the flexibility of Cε of MET146, whereas the rest of the residue fails to adapt being fixed by the chalcogen bond. This effect can be useful to induce selectivity, as the necessary combination of methionine residues only occurs in 9.3% of human kinases, while methionine is the predominant gatekeeper (39%).