Towards holistic Compound Quality Scores: Extending ligand efficiency indices with compound pharmacokinetic characteristics.

The suitability of small molecules as oral drugs is often assessed by simple physicochemical rules, the application of ligand efficiency scores or by composite scores based on physicochemical compound properties. These rules and scores are empirical and typically lack mechanistic background, such as information on pharmacokinetics (PK). We introduce new types of Compound Quality Scores (CQS, specifically called dose scores and cmax scores), which explicitly include predicted or, when available, experimental PK parameters and combine these with on-target potency. These CQS scores are surrogates for an estimated dose and corresponding cmax and allow prioritizing of compounds within test cascades as well as before synthesis. We demonstrate the complementarity and, in most cases, superior performance relative to existing efficiency metrics by project examples.

Rethinking drug design in the artificial intelligence era.

Artificial intelligence (AI) tools are increasingly being applied in drug discovery. While some protagonists point to vast opportunities potentially offered by such tools, others remain sceptical, waiting for a clear impact to be shown in drug discovery projects. The reality is probably somewhere in-between these extremes, yet it is clear that AI is providing new challenges not only for the scientists involved but also for the biopharma industry and its established processes for discovering and developing new medicines. This article presents the views of a diverse group of international experts on the 'grand challenges' in small-molecule drug discovery with AI and the approaches to address them.

Mining electronic laboratory notebooks: analysis, retrosynthesis, and reaction based enumeration.

An approach to automatically analyze and use the knowledge contained in electronic laboratory notebooks (ELNs) has been developed. Reactions were reduced to their reactive center and converted to a string representation (SMIRKS) which formed the basis for reaction classification and in silico (retro-)synthesis. Of the SMIRKS that occurred at least five times, 98% successfully regenerated the original product. The extracted reaction rules (SMIRKS) and corresponding reactants span a virtual chemical space which showed a strong dependence on the size of the reactive center. Whereas relatively few robust reaction types were sufficient to describe a large part of all reactions, considerably more reaction rules were necessary to cover all reactions in the ELN. Furthermore, reaction sequences were extracted to identify frequent combinations and diversifying reaction steps. Based on the extracted knowledge a (retro-)synthesis tool was built allowing for de novo design of compounds which have a high chance of being synthetically accessible. In an example application of the de novo design tool, various feasible retrosynthetic routes to the query molecule were obtained. Reaction based enumeration along the top ranked route yielded a library of 29 920 compounds with diverse properties, 99.9% of which are novel in the sense that they are unknown to the public domain.

A small nonrule of 3 compatible fragment library provides high hit rate of endothiapepsin crystal structures with various fragment chemotypes.

Druglike molecules are defined by Lipinski's rule of 5, to characterize fragment thresholds, they have been reduced from 5 to 3 (Astex's rule of 3). They are applied to assemble fragment libraries, and providers use them to select fragments for commercial offer. We question whether these rules are too stringent to compose fragment libraries with candidates exhibiting sufficient room for chemical subsequent growing and merging modifications as appropriate functional groups for chemical transformations are required. Usually these groups exhibit properties as hydrogen bond donors/acceptors and provide entry points for optimization chemistry. We therefore designed a fragment library (364 entries) without strictly applying the rule of 3. For initial screening for endothiapepsin binding, we performed a biochemical cleavage assay of a fluorogenic substrate at 1 mM. "Hits" were defined to inhibit the enzyme by at least 40%. Fifty-five hits were suggested and subsequently soaked into endothiapepsin crystals. Eleven crystal structures could be determined covering fragments with diverse binding modes: (i) direct binding to the catalytic dyad aspartates, (ii) water-mediated binding to the aspartates, (iii) no direct interaction with the dyad. They occupy different specificity pockets. Only 4 of the 11 fragments are consistent with the rule of 3. Restriction to this rule would have limited the fragment hits to a strongly reduced variety of chemotypes.

Structure and substrate docking of a hydroxy(phenyl)pyruvate reductase from the higher plant Coleus blumei Benth.

Hydroxy(phenyl)pyruvate reductase [H(P)PR] belongs to the family of D-isomer-specific 2-hydroxyacid dehydrogenases and catalyzes the reduction of hydroxyphenylpyruvates as well as hydroxypyruvate and pyruvate to the corresponding lactates. Other non-aromatic substrates are also accepted. NADPH is the preferred cosubstrate. The crystal structure of the enzyme from Coleus blumei (Lamiaceae) has been determined at 1.47 A resolution. In addition to the apoenzyme, the structure of a complex with NADP(+) was determined at a resolution of 2.2 A. H(P)PR is a dimer with a molecular mass of 34 113 Da per subunit. The structure is similar to those of other members of the enzyme family and consists of two domains separated by a deep catalytic cleft. To gain insights into substrate binding, several compounds were docked into the cosubstrate complex structure using the program AutoDock. The results show two possible binding modes with similar docking energy. However, only binding mode A provides the necessary environment in the active centre for hydride and proton transfer during reduction, leading to the formation of the (R)-enantiomer of lactate and/or hydroxyphenyllactate.

Structure-based optimization of aldose reductase inhibitors originating from virtual screening.

Diabetes mellitus is a universal health problem. The World Health Organization (WHO) estimates that 150 million people suffer from diabetes mellitus worldwide in 2005. Long-term complications are a serious problem in the treatment of diabetes, manifesting in macrovascular and microvascular complications. Sorbitol accumulation has been proposed to be an important factor in the development of microvascular complications such as nephropathy, neuropathy, retinopathy or cataract. Catalyzing the NADPH-dependent reduction of glucose to sorbitol, aldose reductase (ALR2) is an important target in the prevention of these complications. The development of novel aldose reductase inhibitors is expected to benefit strongly from a structure-based design approach. A virtual screening based on the ultrahigh-resolution crystal structure of the inhibitor IDD 594 in complex with human ALR2 identified two compounds with IC(50) values in the low micro- to submicromolar range. Based on the known interactions between the ligands and their binding pocket, we simplified the lead structures to give the minimal structural requirements and developed synthetic pathways from commercially available compounds. The newly synthesized compounds were assayed for their inhibition of ALR2, showing inhibitory activities down to the nanomolar range. Crystal structure analysis of the most potent derivative of our series revealed insights into the binding mode of the inhibitors.

Evidence for a novel binding site conformer of aldose reductase in ligand-bound state.

Human aldose reductase (ALR2) has evolved as a promising therapeutic target for the treatment of diabetic long-term complications. The binding site of this enzyme possesses two main subpockets: the catalytic anion-binding site and the hydrophobic specificity pocket. The latter can be observed in the open or closed state, depending on the bound ligand. Thus, it exhibits a pronounced capability for induced-fit adaptations, whereas the catalytic pocket exhibits rigid properties throughout all known crystal structures. Here, we determined two ALR2 crystal structures at 1.55 and 1.65 A resolution, each complexed with an inhibitor of the recently described naphtho[1,2-d]isothiazole acetic acid series. In contrast to the original design hypothesis based on the binding mode of tolrestat (1), both inhibitors leave the specificity pocket in the closed state. Unexpectedly, the more potent ligand (2) extends the catalytic pocket by opening a novel subpocket. Access to this novel subpocket is mainly attributed to the rotation of an indole moiety of Trp 20 by about 35 degrees . The newly formed subpocket provides accommodation of the naphthyl portion of the ligand. The second inhibitor, 3, differs from 2 only by an extended glycolic ester functionality added to one of its carboxylic groups. However, despite this slight structural modification, the binding mode of 3 differs dramatically from that of the first inhibitor, but provokes less pronounced induced-fit adaptations of the binding cavity. Thus, a novel binding site conformation has been identified in a region where previous complex structures suggested only low adaptability of the binding pocket. Furthermore, the two ligand complexes represent an impressive example of how the slight change of a chemically extended side-chain at a given ligand scaffold can result in a dramatically altered binding mode. In addition, our study emphasizes the importance of crystal structure analysis for the translation of affinity data into structure-activity relationships.

Expect the unexpected or caveat for drug designers: multiple structure determinations using aldose reductase crystals treated under varying soaking and co-crystallisation conditions.

In structure-based drug design, accurate crystal structure determination of protein-ligand complexes is of utmost importance in order to elucidate the binding characteristics of a putative lead to a given target. It is the starting point for further design hypotheses to predict novel leads with improved properties. Often, crystal structure determination is regarded as ultimate proof for ligand binding providing detailed insight into the specific binding mode of the ligand to the protein. This widely accepted practise relies on the assumption that the crystal structure of a given protein-ligand complex is unique and independent of the protocol applied to produce the crystals. We present two examples indicating that this assumption is not generally given, even though the composition of the mother liquid for crystallisation was kept unchanged: Multiple crystal structure determinations of aldose reductase complexes obtained under varying crystallisation protocols concerning soaking and crystallisation exposure times were performed resulting in a total of 17 complete data sets and ten refined crystal structures, eight in complex with zopolrestat and two complexed with tolrestat. In the first example, a flip of a peptide bond is observed, obviously depending on the crystallisation protocol with respect to soaking and co-crystallisation conditions. This peptide flip is accompanied by a rupture of an H-bond formed to the bound ligand zopolrestat. The indicated enhanced local mobility of the complex is in agreement with the results of molecular dynamics simulations. As a second example, the aldose reductase-tolrestat complex is studied. Unexpectedly, two structures could be obtained: one with one, and a second with four inhibitor molecules bound to the protein. They are located in and near the binding pocket facilitated by crystal packing effects. Accommodation of the four ligand molecules is accompanied by pronounced shifts concerning two helices interacting with the additional ligands.

High-resolution crystal structure of aldose reductase complexed with the novel sulfonyl-pyridazinone inhibitor exhibiting an alternative active site anchoring group.

The crystal structure of a novel sulfonyl-pyridazinone inhibitor in complex with aldose reductase, the first enzyme of the polyol pathway, has been determined to 1.43 angstroms and 0.95 angstroms resolution. The ternary complex of inhibitor, cofactor and enzyme has been obtained by soaking of preformed crystals. Supposedly due to low solubility in the crystallisation buffer, in both structures the inhibitor shows reduced occupancy of 74% and 46% population, respectively. The pyridazinone head group of the inhibitor occupies the catalytic site, whereas the chloro-benzofuran moiety penetrates into the opened specificity pocket. The high-resolution structure provides some evidence that the pyridazinone group binds in a negatively charged deprotonated state, whereas the neighbouring His110 residue most likely adopts a neutral uncharged status. Since the latter structure is populated by the ligand to only 46%, a second conformation of the C-terminal ligand-binding region can be detected. This conformation corresponds to the closed state of the specificity pocket when no or only small ligands are bound to aldose reductase. The two conformational states are in good agreement with frames observed along a molecular dynamics trajectory describing the transition from closed to open situation. Accordingly, both geometries, superimposed in the averaged crystal structure, correspond to snapshots of the ligand-bound and the unbound state. Isothermal titration calorimetry has been applied to determine the binding constants of the investigated pyridazinone in comparison to the hydantoin sorbinil and the carboxylate-type inhibitors IDD 594 and tolrestat. The pyridazinone exhibits a binding affinity similar to those of tolrestat and sorbinil, and shows slightly reduced affinity compared to IDD 594. These studies elucidating the binding mode and providing information about protonation states of protein side-chains involved in binding of this novel class of inhibitors establish the platform for further structure-based drug design.