Discovery and Optimization of Potent, Efficacious and Selective Inhibitors Targeting EGFR Exon20 Insertion Mutations

Herein, we report the identification and optimization of a series of potent inhibitors of EGFR Exon20 insertions with significant selectivity over wild-type EGFR. A strategically designed HTS campaign, multiple iterations of structure-based drug design (SBDD), and tactical linker replacement led to a potent and wild-type selective series of molecules and ultimately the discovery of 36. Compound 36 is a potent and selective inhibitor of EGFR Exon20 insertions and has demonstrated encouraging efficacy in NSCLC EGFR CRISPR-engineered H2073 xenografts that carry an SVD Exon20 insertion and reduced efficacy in a H2073 wild-type EGFR xenograft model compared to CLN-081 (5), indicating that 36 may have lower EGFR wild-type associated toxicity.

Discovery and characterisation of quinazolines and 8-Azaquinazolines as NLRP3 agonists with oral bioavailability in mice

The NLRP3 inflammasome is a multiprotein complex that plays a critical role in activating the immune system in response to danger signals. Small molecule agonists of NLRP3 may offer clinical benefits in cancer immunology either as a monotherapy or in combination with checkpoint blockade, where it is hypothesised that their application can help to initiate an antitumor immune response. In this study, we report the discovery of quinazolines and 8-azaquinazolines as NLRP3 agonists and their chemical optimization to afford compounds with oral bioavailability in mice. We confirm that these compounds engage the NLRP3 inflammasome by verifying their dependence upon lipopolysaccharide (LPS) priming for cytokine release and the activation of Caspase-1. We further demonstrate pathway engagement through loss of activity in an NLRP3-knockout THP1 cell line. Based on their pharmacokinetic profile and biological activity, these compounds represent valuable tools to evaluate the therapeutic potential of NLRP3 activation in a pre-clinical setting.

Abstract 5076: Evaluation of the CNS penetration of a next generation PARP inhibitor, AZD9574, in cynomolgus monkey using positron emission tomography

The current clinically approved PARP inhibitors have limited subtype selectivity and are to some degree restricted in their ability to penetrate the central nervous system (CNS) due to efflux transporters, potentially limiting their efficacy in treating metastatic disease or primary tumors in the brain. The current study evaluated the potential of AZD9574, a next generation, PARP1 selective inhibitor/trapper, to penetrate the CNS in cynomolgus monkey, and its occupancy of the PARP1 enzyme, using positron emission tomography (PET). In vitro bidirectional efflux assay data suggested AZD9574 showed minimal substrate potential compared to the clinically approved PARP inhibitors. This was reflected in an increased ratio of unbound brain to unbound plasma concentration (Kpuu) in the rat of ~0.31. Therefore AZD9574 was taken forward into cynomolgus monkey PET studies. Firstly, the Kpuu was determined following dosing of [11C]AZD9574, co-administered with unlabeled drug to minimize the impact of specific binding. The high specific signal observed lead to the development of [11C]AZ3391, a PARP1 selective, CNS penetrant PET tracer, which was subsequently used to directly assess the PARP1 target engagement of AZD9574 in the brain. AZD9574 was found to show a Kpuu in cynomolgus monkeys of 0.79, close to unity with unbound plasma concentrations suggesting minimal CNS restriction. Furthermore, an i.v. infusion dose response study with AZD9574, conducted to examine its ability to block target occupancy by the PET tracer [11C]AZ3391, demonstrated a reduction in [11C]AZ3391 accumulation in whole brain. The resulting calculated occupancy of AZD9574 ranged from 17% for the lowest dose (0.003 mg/kg) to 95% for the highest dose tested (0.05 mg/kg). Comparable reduction in occupancy was seen for peripheral tissue, such as bone marrow, supporting the conclusion that AZD9574 shows minimal CNS restriction. These data show that AZD9574 is the first PARP inhibitor to reach the clinic which combines PARP1 selectivity, trapping and high CNS penetration in a single molecule and supports its development as a potential therapy for the treatment of metastatic disease and primary brain tumors.

Citation Format: Andy Pike, Amber Balazs, Zsolt Cselényi, Sébastien L. Degorce, Avipsa Ghosh, Sudhir M. Hande, Jeffrey Johannes, Peter Johnström, Martin J. Packer, Magnus Schou, XiaoLan Zheng. Evaluation of the CNS penetration of a next generation PARP inhibitor, AZD9574, in cynomolgus monkey using positron emission tomography [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5076.

NMR Molecular Replacement Provides New Insights into Binding Modes to Bromodomains of BRD4 and TRIM24

Structure-based drug discovery (SBDD) largely relies on structural information from X-ray crystallography because traditional NMR structure calculation methods are too time consuming to be aligned with typical drug discovery timelines. The recently developed NMR molecular replacement (NMR²) method dramatically reduces the time needed to generate ligand-protein complex structures using published structures (apo or holo) of the target protein and treating all observed NOEs as ambiguous restraints, bypassing the laborious process of obtaining sequence-specific resonance assignments for the protein target. We apply this method to two therapeutic targets, the bromodomain of TRIM24 and the second bromodomain of BRD4. We show that the NMR² methodology can guide SBDD by rationalizing the observed SAR. We also demonstrate that new types of restraints and selective methyl labeling have the potential to dramatically reduce "time to structure" and extend the method to targets beyond the reach of traditional NMR structure elucidation.

Discovery of 5-{4-[(7-Ethyl-6-oxo-5,6-dihydro-1,5-naphthyridin-3-yl)methyl]piperazin-1-yl}-N-methylpyridine-2-carboxamide (AZD5305): A PARP1-DNA Trapper with High Selectivity for PARP1 over PARP2 and Other PARPs

Poly-ADP-ribose-polymerase (PARP) inhibitors have achieved regulatory approval in oncology for homologous recombination repair deficient tumors including BRCA mutation. However, some have failed in combination with first-line chemotherapies, usually due to overlapping hematological toxicities. Currently approved PARP inhibitors lack selectivity for PARP1 over PARP2 and some other 16 PARP family members, and we hypothesized that this could contribute to toxicity. Recent literature has demonstrated that PARP1 inhibition and PARP1-DNA trapping are key for driving efficacy in a BRCA mutant background. Herein, we describe the structure- and property-based design of 25 (AZD5305), a potent and selective PARP1 inhibitor and PARP1-DNA trapper with excellent in vivo efficacy in a BRCA mutant HBCx-17 PDX model. Compound 25 is highly selective for PARP1 over other PARP family members, with good secondary pharmacology and physicochemical properties and excellent pharmacokinetics in preclinical species, with reduced effects on human bone marrow progenitor cells in vitro.

Discovery and pharmacological characterization of AZD3229, a potent KIT/PDGFRα inhibitor for treatment of gastrointestinal stromal tumors

Gastrointestinal stromal tumor (GIST) is the most common human sarcoma driven by mutations in KIT or platelet-derived growth factor α (PDGFRα). Although first-line treatment, imatinib, has revolutionized GIST treatment, drug resistance due to acquisition of secondary KIT/PDGFRα mutations develops in a majority of patients. Second- and third-line treatments, sunitinib and regorafenib, lack activity against a plethora of mutations in KIT/PDGFRα in GIST, with median time to disease progression of 4 to 6 months and inhibition of vascular endothelial growth factor receptor 2 (VEGFR2) causing high-grade hypertension. Patients with GIST have an unmet need for a well-tolerated drug that robustly inhibits a range of KIT/PDGFRα mutations. Here, we report the discovery and pharmacological characterization of AZD3229, a potent and selective small-molecule inhibitor of KIT and PDGFRα designed to inhibit a broad range of primary and imatinib-resistant secondary mutations seen in GIST. In engineered and GIST-derived cell lines, AZD3229 is 15 to 60 times more potent than imatinib in inhibiting KIT primary mutations and has low nanomolar activity against a wide spectrum of secondary mutations. AZD3229 causes durable inhibition of KIT signaling in patient-derived xenograft (PDX) models of GIST, leading to tumor regressions at doses that showed no changes in arterial blood pressure (BP) in rat telemetry studies. AZD3229 has a superior potency and selectivity profile to standard of care (SoC) agents-imatinib, sunitinib, and regorafenib, as well as investigational agents, avapritinib (BLU-285) and ripretinib (DCC-2618). AZD3229 has the potential to be a best-in-class inhibitor for clinically relevant KIT/PDGFRα mutations in GIST.

IMPRESSION - prediction of NMR parameters for 3-dimensional chemical structures using machine learning with near quantum chemical accuracy

The IMPRESSION (Intelligent Machine PREdiction of Shift and Scalar information Of Nuclei) machine learning system provides an efficient and accurate method for the prediction of NMR parameters from 3-dimensional molecular structures. Here we demonstrate that machine learning predictions of NMR parameters, trained on quantum chemical computed values, can be as accurate as, but computationally much more efficient (tens of milliseconds per molecular structure) than, quantum chemical calculations (hours/days per molecular structure) starting from the same 3-dimensional structure. Training the machine learning system on quantum chemical predictions, rather than experimental data, circumvents the need for the existence of large, structurally diverse, error-free experimental databases and makes IMPRESSION applicable to solving 3-dimensional problems such as molecular conformation and stereoisomerism.

Utilizing Grand Canonical Monte Carlo Methods in Drug Discovery

The concepts behind targeting waters for potency and selectivity gains have been well documented and explored, although maximizing such potential gains can prove to be challenging. This problem is exacerbated in cases where there are multiple interacting waters, wherein perturbation of one water can affect the free energy landscape of the remaining waters. Knowing the right modification a priori is challenging, and computational approaches are ideally suited to help answer the key question of which substitution is best to try. Here, we use Grand Canonical Monte Carlo and the recent Grand Canonical Alchemical Perturbation methods to both understand and predict the effect of ligand-mediated water displacement when more than one water molecule is involved, as well as to understand how exploiting water networks can help govern selectivity.

Free Ligand 1D NMR Conformational Signatures To Enhance Structure Based Drug Design of a Mcl-1 Inhibitor (AZD5991) and Other Synthetic Macrocycles

The three-dimensional conformations adopted by a free ligand in solution impact bioactivity and physicochemical properties. Solution 1D NMR spectra inherently contain information on ligand conformational flexibility and three-dimensional shape, as well as the propensity of the free ligand to fully preorganize into the bioactive conformation. Herein we discuss some key learnings, distilled from our experience developing potent and selective synthetic macrocyclic inhibitors, including Mcl-1 clinical candidate AZD5991. Case studies have been selected from recent oncology research projects, demonstrating how 1D NMR conformational signatures can complement X-ray protein-ligand structural information to guide medicinal chemistry optimization. Learning to extract free ligand conformational information from routinely available 1D NMR signatures has proven to be fast enough to guide medicinal chemistry decisions within design cycles for compound optimization.

Discovery of N-(4-{[5-Fluoro-7-(2-methoxyethoxy)quinazolin-4-yl]amino}phenyl)-2-[4-(propan-2-yl)-1 H-1,2,3-triazol-1-yl]acetamide (AZD3229), a Potent Pan-KIT Mutant Inhibitor for the Treatment of Gastrointestinal Stromal Tumors

While the treatment of gastrointestinal stromal tumors (GISTs) has been revolutionized by the application of targeted tyrosine kinase inhibitors capable of inhibiting KIT-driven proliferation, diverse mutations to this kinase drive resistance to established therapies. Here we describe the identification of potent pan-KIT mutant kinase inhibitors that can be dosed without being limited by the tolerability issues seen with multitargeted agents. This effort focused on identification and optimization of an existing kinase scaffold through the use of structure-based design. Starting from a series of previously reported phenoxyquinazoline and quinoline based inhibitors of the tyrosine kinase PDGFRα, potency against a diverse panel of mutant KIT driven Ba/F3 cell lines was optimized, with a particular focus on reducing activity against a KDR driven cell model in order to limit the potential for hypertension commonly seen in second and third line GIST therapies. AZD3229 demonstrates potent single digit nM growth inhibition across a broad cell panel, with good margin to KDR-driven effects. Selectivity over KDR can be rationalized predominantly by the interaction of water molecules with the protein and ligand in the active site, and its kinome selectivity is similar to the best of the approved GIST agents. This compound demonstrates excellent cross-species pharmacokinetics, shows strong pharmacodynamic inhibition of target, and is active in several in vivo models of GIST.

Correction to "Structure Based Design of Non-Natural Peptidic Macrocyclic Mcl-1 Inhibitors"

[This corrects the article DOI: 10.1021/acsmedchemlett.6b00464.].

Structure Based Design of Non-Natural Peptidic Macrocyclic Mcl-1 Inhibitors

Mcl-1 is a pro-apoptotic BH3 protein family member similar to Bcl-2 and Bcl-xL. Overexpression of Mcl-1 is often seen in various tumors and allows cancer cells to evade apoptosis. Here we report the discovery and optimization of a series of non-natural peptide Mcl-1 inhibitors. Screening of DNA-encoded libraries resulted in hit compound 1, a 1.5 μM Mcl-1 inhibitor. A subsequent crystal structure demonstrated that compound 1 bound to Mcl-1 in a β-turn conformation, such that the two ends of the peptide were close together. This proximity allowed for the linking of the two ends of the peptide to form a macrocycle. Macrocyclization resulted in an approximately 10-fold improvement in binding potency. Further exploration of a key hydrophobic interaction with Mcl-1 protein and also with the moiety that engages Arg256 led to additional potency improvements. The use of protein-ligand crystal structures and binding kinetics contributed to the design and understanding of the potency gains. Optimized compound 26 is a <3 nM Mcl-1 inhibitor, while inhibiting Bcl-2 at only 5 μM and Bcl-xL at >99 μM, and induces cleaved caspase-3 in MV4-11 cells with an IC₅₀ of 3 μM after 6 h.

Predicting the relative binding affinity of mineralocorticoid receptor antagonists by density functional methods

In drug discovery, prediction of binding affinity ahead of synthesis to aid compound prioritization is still hampered by the low throughput of the more accurate methods and the lack of general pertinence of one method that fits all systems. Here we show the applicability of a method based on density functional theory using core fragments and a protein model with only the first shell residues surrounding the core, to predict relative binding affinity of a matched series of mineralocorticoid receptor (MR) antagonists. Antagonists of MR are used for treatment of chronic heart failure and hypertension. Marketed MR antagonists, spironolactone and eplerenone, are also believed to be highly efficacious in treatment of chronic kidney disease in diabetes patients, but is contra-indicated due to the increased risk for hyperkalemia. These findings and a significant unmet medical need among patients with chronic kidney disease continues to stimulate efforts in the discovery of new MR antagonist with maintained efficacy but low or no risk for hyperkalemia. Applied on a matched series of MR antagonists the quantum mechanical based method gave an R(2) = 0.76 for the experimental lipophilic ligand efficiency versus relative predicted binding affinity calculated with the M06-2X functional in gas phase and an R(2) = 0.64 for experimental binding affinity versus relative predicted binding affinity calculated with the M06-2X functional including an implicit solvation model. The quantum mechanical approach using core fragments was compared to free energy perturbation calculations using the full sized compound structures.

The Dalton quantum chemistry program system

Dalton is a powerful general-purpose program system for the study of molecular electronic structure at the Hartree-Fock, Kohn-Sham, multiconfigurational self-consistent-field, Møller-Plesset, configuration-interaction, and coupled-cluster levels of theory. Apart from the total energy, a wide variety of molecular properties may be calculated using these electronic-structure models. Molecular gradients and Hessians are available for geometry optimizations, molecular dynamics, and vibrational studies, whereas magnetic resonance and optical activity can be studied in a gauge-origin-invariant manner. Frequency-dependent molecular properties can be calculated using linear, quadratic, and cubic response theory. A large number of singlet and triplet perturbation operators are available for the study of one-, two-, and three-photon processes. Environmental effects may be included using various dielectric-medium and quantum-mechanics/molecular-mechanics models. Large molecules may be studied using linear-scaling and massively parallel algorithms. Dalton is distributed at no cost from http://www.daltonprogram.org for a number of UNIX platforms.

Pharmacophore binding motifs for nicotinamide adenine dinucleotide analogues across multiple protein families: a detailed contact-based analysis of the interaction between proteins and NAD(P) cofactors

We have analyzed the protein-binding pharmacophore of NAD and its close analogues in all protein-ligand structures available in the RCSB database as of February 2012; this analysis has then been used to assess the novelty of structures emerging after that date. We show that proteins have evolved diverse pharmacophore motifs for binding the adenine moiety, fewer, but still diverse, motifs for nicotinamide, and a very limited set of motifs for binding the pyrophosphate linker. Our exhaustive analysis includes a pharmacophore contact analysis for over 1900 protein-ligand structures containing NAD analogues; we have benchmarked this set of contacts against nearly 27 000 protein-ligand structures to demonstrate that the diversity of interactions seen with NAD is very similar to that seen for all other ligands. Hence, variation in binding motifs for NAD is not distinct from that observed for other ligands and they show significant variation across protein families.

Quantification of free ligand conformational preferences by NMR and their relationship to the bioactive conformation

Accurate unbound solution 3D-structures of ligands provide unique opportunities for medicinal chemistry and, in particular, a context to understand binding thermodynamics and kinetics. Previous methods of deriving these 3D-structures have had neither the accuracy nor resolution needed for drug design and have not yet realized their potential. Here, we describe and apply a NMR methodology to the aminoglycoside streptomycin that can accurately quantify accessible 3D-space and rank the occupancy of observed conformers to a resolution that enables medicinal chemistry understanding and design. Importantly, it is based upon conventional small molecule NMR techniques and can be performed in physiologically-relevant solvents. The methodology uses multiple datasets, an order of magnitude more experimental data than previous NMR approaches and a dynamic model during refinement, is independent of computational chemistry and avoids the problem of virtual conformations. The refined set of solution 3D-shapes for streptomycin can be grouped into two major families, of which the most populated is almost identical to the 30S ribosomal subunit bioactive shape. We therefore propose that accurate unbound ligand solution conformations may, in some cases, provide a subsidiary route to bioactive shape without crystallography. This experimental technique opens up new opportunities for drug design and more so when complemented with protein co-crystal structures, SAR data and pharmacophore modeling.

Discovery of a potent, selective, and orally bioavailable acidic 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1) inhibitor: discovery of 2-[(3S)-1-[5-(cyclohexylcarbamoyl)-6-propylsulfanylpyridin-2-yl]-3-piperidyl]acetic acid (AZD4017)

Inhibition of 11β-HSD1 is an attractive mechanism for the treatment of obesity and other elements of the metabolic syndrome. We report here the discovery of a nicotinic amide derived carboxylic acid class of inhibitors that has good potency, selectivity, and pharmacokinetic characteristics. Compound 11i (AZD4017) is an effective inhibitor of 11β-HSD1 in human adipocytes and exhibits good druglike properties and as a consequence was selected for clinical development.

Adipose tissue-derived adiponectin expression is significantly associated with increased post operative mortality in horses undergoing emergency abdominal surgery

REASONS FOR PERFORMING STUDY

Adipose tissue is an important source of inflammatory cytokines (adipokines) and adiposity has been identified as having a significant effect on human morbidity and mortality. Obesity is also an emerging welfare problem in the UK horse population, but the role that it plays in secondary diseases is unclear.

OBJECTIVES

To examine the expression of inflammation-related adipokine genes in retroperitoneal adipose tissue of horses undergoing emergency abdominal surgery and to explore associations with adiposity and post operative survival.

METHODS

Retroperitoneal adipose tissue samples were obtained from 76 horses undergoing emergency abdominal surgery. Real-time PCR was used to measure gene expression for leptin, adiponectin, tumour necrosis factor-alpha, macrophage chemoattractant protein-1, macrophage inhibitory factor, serum amyloid A, haptoglobin and interleukin-1. Multivariate patterns of adipokine expression were explored with principal component analysis (PCA), whilst univariable associations with post operative survival were tested in a Cox proportional hazards model.

RESULTS

Leptin gene expression was higher in overweight and obese horses than in lean animals. Expression of mRNA encoding adiponectin mRNA in visceral adipose tissue was positively associated with increased post operative mortality (hazard ratio 1.31, 95% CI 1.05-1.65). However, PCA did not demonstrate multivariable patterns of adipokine gene expression from visceral adipose tissue associated with body mass index or with survival.

CONCLUSIONS

In horses presented with acute intestinal disease, increased adiponectin gene expression from retroperitoneal adipose tissue is associated with an increased risk of mortality. Obesity assessed by BMI had no association with increased post operative mortality in horses with primary gastrointestinal disease.

POTENTIAL RELEVANCE

Further study is warranted on the expression and effects of adipokines, particularly adiponectin, and correlation with postoperative outcome.

Evidence for partially bound states in cooperative molecular recognition interfaces

A zinc porphyrin equipped with four amide H-bonding sites provides a rigid molecular receptor for the study of cooperative multipoint binding interactions. The interaction of this receptor with a variety of pyridine ligands bearing zero, one, and two H-bonding sites has been studied using UV/vis absorption, (1)H and (31)P NMR spectroscopy, and isothermal titration calorimetry in five different solvents. The results are analyzed in terms of a bound state that populates an ensemble of different complexes in which zero, one, or two of the potential H-bond interactions are formed. The key parameter that determines the behavior of the system is the product of the association constant for the H-bond interaction, K(H), and the effective molarity for the intramolecular interaction, EM. In the system reported here, EM is 0.1-1 M for all of the intramolecular interactions. For strong H-bonds (large K(H) in nonpolar solvents), all of the interactions are formed in the complex and the fully bound state dominates. In this case, additional binding interactions produce incremental increases in complex stability. However, for weaker H-bonds (small K(H) in polar solvents), the formation of additional interactions does not lead to an increase in the overall stability of the complex, due to the population of partially bound states.

Use of quantitative (1)H NMR chemical shift changes for ligand docking into barnase

(1)H NMR complexation-induced changes in chemical shift (CIS) of HN protons have been used to characterize the complexes of barnase with the deoxyoligonucleotides d(GC) and d(CGAC). Quantitative shift changes are used not only to locate the most probable binding site (using ring-current shifts), but also to determine the orientation of the ligand within the binding site, based on a more complete shift calculation including bond magnetic anisotropies and electric field effects. For both ligands, the guanine is in the same binding site cleft, in the same position as identified in the crystal structure of the d(CGAC) complex. By contrast, a previous X-ray crystal structure of the d(GC) complex showed the ligand in the mouth of the active site, rather than at the guanyl-specific site, implying that the location may be an artifact of the crystallisation process.

Influence of conformational flexibility on complexation-induced changes in chemical shift in a neocarzinostatin protein-ligand complex

In this paper is described an analysis of the effects of protein flexibility on the observed CIS values and the impact on the accuracy of 3D structures determined using a (1)H NMR CIS approach. The effects of protein conformational mobility have been investigated by using a set of different protein structures as starting points for the calculation: the unbound X-ray crystal structure, the unbound NMR solution structure, and the bound NMR solution structure of the protein. The results indicated that loop movement does have a significant impact on the quality of the structure generated by the CIS structure determination methodology. The implementation of methods to treat loop flexibility within our protocol, however, did not improve the results for calculations based on the unbound protein frame.

Determination of protein-ligand binding modes using complexation-induced changes in (1)h NMR chemical shift

A new method for determining three-dimensional solution structures of protein-ligand complexes using experimentally determined complexation-induced changes in (1)H NMR chemical shift (CIS) is introduced. The method has been validated using the complex formed between the protein antitumor antibiotic neocarzinostatin (NCS) and a synthetic chromophore analogue. The X-ray crystal structure of the unbound protein and the backbone amide proton CIS were the input data used in the determination of the three-dimensional structure of the complex. The experimental CIS values were used in a continuous direct structure refinement process based on genetic algorithms to sample conformational space. The calculated structure of the complex agrees well with the NMR solution structure, indicating the potential of this approach for structure determination.

PREDICTOR: a web-based tool for the prediction of atomic structure from sequence for double helical DNA with up to 150 base pairs

We announce the release of a web-based tool for DNA structure prediction (PREDICTOR) which allows the calculation of atomic structures of double-helical DNA with up to 150 Watson-Crick base pairs (http://farwer.staff.shef.ac.uk/ PREDICTOR). The semi-empirical method uses computational chemistry to extrapolate knowledge of sequence-dependent DNA structure contained in the X-ray crystal structure database. The properties of the base stacking interactions are treated theoretically, and an empirical model is used to add the conformational constraints imposed by the backbone. For DNA oligomers in the X-ray crystal structure database that were not used for parameterisation of the model, the method distinguishes A and B form DNA reasonably reliably, and the final structures are accurate to 2 å rmsd. Simulation of a 150mer and a 494mer with experimentally confirmed bending clearly reproduces the bending whereas the predicted structure of a random 150mer does not show any curvature. Calculation times are 90 seconds for an octamer and 7 minutes for a 30mer.

Prediction of atomic structure from sequence for double helical DNA oligomers

DNA can adopt different conformations depending on the base sequence, solvent, electrolyte composition and concentration, pH, temperature, and interaction with proteins. Here we present a model for calculating the three-dimensional atomic structure of double-stranded DNA oligomers. A theoretical energy function is used for calculating the interactions within the base steps and an empirical backbone function is used to restrict the conformational space accessible to the bases and to account for the conformational coupling of neighboring steps in a sequence. Conformational searching on large structures or a large number of structures is possible, because each base step can be described by just two primary degrees of freedom (slide and shift). A genetic algorithm is used to search for low-energy structures in slide-shift space, and this allows very rapid optimization of DNA oligomers. The other base step parameters have been previously optimized for all possible slide-shift sequence combinations, and a heuristic algorithm is used to add the atomic details of the backbone conformation in the final step of the calculation. The structures obtained by this method are very similar to the corresponding X-ray crystal structures observed experimentally. The average RMSD is 2.24 Angstroms for a set of 20 oligomer structures. For 15 of these sequences, the X-ray crystal structure is the global energy minimum. The other 5 are bistable sequences that have B-form global energy minima but crystallize as A-DNA.

Sequence-dependent DNA structure: a database of octamer structural parameters

We have constructed the potential energy surfaces for all unique tetramers, hexamers and octamers in double helical DNA, as a function of the two principal degrees of freedom, slide and shift at the central step. From these potential energy maps, we have calculated a database of structural and flexibility properties for each of these sequences. These properties include: the values of each of the six step parameters (twist roll, tilt, rise, slide and shift), for each step of the sequence; flexibility measures for both decrease and increase in each property value from the minimum energy conformation for the central step; and the deviation from the path of a hypothetical straight octamer. In an analysis of structural change as a function of sequence length, we observe that almost all DNA tends to B-DNA and becomes less flexible. A more detailed analysis of octamer properties has allowed us to determine the structural preferences of particular sequence elements. GGC and GCC sequences tend to confer bistability, low stability and a predisposition to A-form DNA, whereas AA steps strongly prefer B-DNA and inhibit A-structures. There is no correlation between flexibility and intrinsic curvature, but bent DNA is less stable than straight. The most difficult deformation is undertwisting. The TA step stands out as the most flexible sequence element with respect to decreasing twist and increasing roll. However, as with the structural properties, this behavior is highly context-dependent and some TA steps are very straight.

Complexation-induced chemical shifts--ab initio parameterization of transferable bond anisotropies

Complexation-induced changes in proton chemical shifts provide a potent tool for conformational analysis, being highly dependent on intermolecular orientation. An important contribution to these shifts arises from the molecular magnetisability anisotropy, or more specifically from the anisotropy of certain groups, such as aromatic rings and unsaturated bonds. While the influence of aromatic rings has been well characterised via the ring current effect, unsaturated bonds have received much less attention and prediction of complexation shifts is hampered by the lack of accurate anisotropy parameters for these bonds. We have therefore used ab initio calculations at the HF/aug-cc-pVDZ level to obtain bond anisotropies for C-H, N-H, C=O, C=C, C triple bond N, N=N, C triple bond C, and C triple bond N. Fitting the anisotropies to bond magnetic dipoles (the McConnell equation) gives non-transferable values for C-H and N-H bonds. We have therefore expanded in terms of bond magnetic dipoles, quadrupoles, and octopoles for double and triple bonds only, obtaining highly accurate shielding surfaces in all cases. The transferable nature of the anisotropies is confirmed by comparing with shifts obtained in larger molecules containing unsaturated bonds.

An evaluation of force-field treatments of aromatic interactions

Experimental measurements of edge-to-face aromatic interactions have been used to test a series of molecular mechanics force fields. The experimental data were determined for a range of differently substituted aromatic rings using chemical double mutant cycles on hydrogen-bonded zipper complexes. These complexes were truncated for the purposes of the molecular mechanics calculations so that problems of conformational searching and the optimisation of large structures could be avoided. Double-mutant cycles were then carried out in silico using these truncated systems. Comparison of the experimental aromatic interaction energies and the X-ray crystal structures of these truncated complexes with the calculated data show that conventional molecular mechanics force fields (MM2, MM3, AMBER and OPLS) do not perform well. However, the XED force field which explicitly represents electron anisotropy as an expansion of point charges around each atom reproduces the trends in interaction energy and the three-dimensional structures exceedingly well. Collapsing the XED charges onto atom centres or the use of semi-empirical atom-centred charges within the XED force field gives poor results. Thus the success of XED is not related to the methods used to assign the atomic charge distribution but can be directly attributed to the use of off-atom centre charges.

Sequence-structure relationships in DNA oligomers: a computational approach

A collective-variable model for DNA structure is used to predict the conformation of a set of 30 octamer, decamer, and dodecamer oligomers for which high-resolution crystal structures are available. The model combines an all-atom base pair representation with an empirical backbone, emphasizing the role of base stacking in fixing sequence-dependent structure. We are able to reproduce trends in roll and twist to within 5 degrees across a large database of both A- and B-DNA oligomers. A genetic algorithm approach is used to search for global minimum structures and this is augmented by a grid search to identify local minimums. We find that the number of local minimums is highly sequence dependent, with certain sequences having a set of minimums that span the entire range between canonical A- and B-DNA conformations. Although the global minimum does not always agree with the crystal structure, for 24 of the 30 oligomers, we find low-energy local minimums that match the experimental step parameters. Discrepancies throw some light on the role of crystal packing in determining the solid-state conformation of double-helical DNA.

Sequence-dependent DNA structure: dinucleotide conformational maps

We have used a computational model to calculate the potential energy surface for dinucleotide steps in double helical DNA as a function of the two principal degrees of freedom, slide and shift. By using a virtual bond to model the constraints imposed by the sugar-phosphate backbone, twist, roll, tilt and rise can be simultaneously optimised for any given values of slide and shift. Thus we have been able to construct complete conformational maps for all step types. For some steps, the maps agree well with experimental data from X-ray crystal structures, but other steps appear to be strongly perturbed by the effects of context (conformational coupling with the neighbouring steps). The optimised values of twist and roll show sequence-dependent variations consistent with the crystal structure data. The conformational maps allow us to construct adiabatic paths, and hence calculate the flexibility of each step with respect to slide and shift. Again the results agree well with the available experimental assignments of flexibility: YR steps, CA/TG and CG, are the most flexible and RR steps, such as AA, the least flexible.

Sequence-dependent DNA structure: tetranucleotide conformational maps

A database of X-ray crystal structures of double helical DNA oligomers has been used to analyse the role of the sugar-phosphate backbone in coupling the conformational properties of neighbouring dinucleotide steps. The base step parameters which are most strongly coupled to the backbone degrees of freedom are slide and shift, and these are the two dinucleotide step parameters which show strong correlations along a sequence: the value of slide follows the values in the neighbouring steps, whereas shift tends to alternate. This conformational coupling is mediated by the shared furanose rings at the step junctions: a change in the value of slide causes a change in the mean value of the same strand 3' and 5'-chi torsion angle, and a change in the mean value of the 3' and 5' sugar pseudo-rotation phase angle, P; a change in the value of shift causes a difference between the same strand 3' and 5'-chi in A-DNA and a difference between the 3' and 5'-P in B-DNA. We have used a database of tetranucleotide X-ray crystal structures to parameterise a simple model for the coupling of slide and shift. Using this junction model together with our dinucleotide step potential energy maps described previously, we can in principle calculate the structure of any DNA oligomer. The parameterisation indicates that the rotational step parameters are accurate to within 5 degrees, and the translational step parameters are accurate to within 0.5 A. The model has been used to study the potential energy surfaces of all possible tetranucleotide sequences, and the calculations agree well with the experimental data from X-ray crystal structures. Some dinucleotide steps are context independent (AA/TT, AT and TA), because the conformational properties of all possible neighbouring steps are compatible. When the conformational properties of the neighbours are not compatible, the behaviour of a step cannot be understood at the dinucleotide level. Thus the conformations of CG, GC and GG/CC are all strongly context dependent. The remaining mixed sequence steps show weakly context-dependent behaviour. The approach allows the calculation of the relative stability and flexibility of tetranucleotide sequences, and the results indicate why TATA is used as an origin of replication. Clear predictions are made about sequences which have not yet been characterised crystallographically. In particular, poly(CCA).poly(TGG) is predicted to have an unusual structure which lies between the C and D-DNA polymorphs.

Sequence-dependent DNA structure: the role of the sugar-phosphate backbone

A detailed analysis of the coupling between the conformational properties of the sugar-phosphate backbone and the base stacking interactions in dinucleotide steps of double helical DNA is described. In X-ray crystal structures of oligonucleotides, the backbone shows one major degree of freedom, consisting of the torsion angles chi, delta, zeta and the pseudorotation phase angle, P. The remaining torsion angles (beta, epsilon, alpha and gamma) comprise two less important degrees of freedom. The base stacking interactions show three degrees of freedom: slide-roll-twist, shift-tilt, and rise (which is more or less constant). Coupling is observed between the base and backbone degrees of freedom. The major base stacking mode, slide-roll-twist, is coupled to the major backbone mode, chi-P-delta-zeta. The secondary base stacking mode, shift-tilt, is coupled to epsilon and zeta and to a lesser extent to the chi-P-delta-zeta mode. We show that the length of the backbone, C, given by the same strand C1'-C1' separation, is an excellent single parameter descriptor for the conformation of the backbone and the way in which it is coupled to the base stacking geometry. The slide-roll-twist motion relates to changes in the mean backbone length, C, and the shift-tilt motion to the difference between the lengths of the two backbone strands, DeltaC. We use this observation to develop a simple virtual bond model which describes the coupling of the backbone conformations and the base stacking geometry. A semi-flexible bond is used to connect the same strand C1'-C1' atoms. Analysis of the X-ray crystal structure database, simple geometric considerations and model building experiments all show that this bond is flexible with respect to slide, shift and propeller but rigid with respect to the other 14 local base stacking parameters. Using this simple model for the backbone in conjunction with potential energy calculations of the base stacking interactions, we show that it is possible to predict accurately the values of these 14 base step parameters, given values of slide, shift and propeller. We also show that the base step parameters fall into three distinct groups: roll, tilt and rise are determined solely by the base stacking interactions and are independent of the backbone; twist is insensitive to the base stacking interactions and is determined solely by the constraints of a relatively rigid fixed length backbone; slide and shift are the primary degrees of freedom and cannot be predicted accurately at the dinucleotide level because they are influenced by the conformations of neighbouring steps in a sequence. We have found that the context effect on slide is mediated by the chi torsion angles while the context effect on shift results from a BI to BII transition in the backbone. We have therefore reduced the dimensionality of the dinucleotide step problem to two parameters, slide and shift.