Andreas Labahn

Towards accurate free energy calculations in ligand protein-binding studies.

Cells contain a multitude of different chemical reaction paths running simultaneously and quite independently next to each other. This amazing feat is enabled by molecular recognition, the ability of biomolecules to form stable and specific complexes with each other and with their substrates. A better understanding of this process, i.e. of the kinetics, structures and thermodynamic properties of biomolecule binding, would be invaluable in the study of biological systems. In addition, as the mode of action of many pharmaceuticals is based upon their inhibition or activation of biomolecule targets, predictive models of small molecule receptor binding are very helpful tools in rational drug design. Since the goal here is normally to design a new compound with a high inhibition strength, one of the most important thermodynamic properties is the binding free energy DeltaG(0). The prediction of binding constants has always been one of the major goals in the field of computational chemistry, because the ability to reliably assess a hypothetical compounds binding properties without having to synthesize it first would save a tremendous amount of work. The different approaches to this question range from fast and simple empirical descriptor methods to elaborate simulation protocols aimed at putting the computation of free energies onto a solid foundation of statistical thermodynamics. While the later methods are still not suited for the screenings of thousands of compounds that are routinely performed in computational drug design studies, they are increasingly put to use for the detailed study of protein ligand interactions. This review will focus on molecular mechanics force field based free energy calculations and their application to the study of protein ligand interactions. After a brief overview of other popular methods for the calculation of free energies, we will describe recent advances in methodology and a variety of exemplary studies of molecular dynamics simulation based free energy calculations.

CONFORMATIONAL CHANGES OF THE H+-ATPASE FROM ESCHERICHIA COLI UPON NUCLEOTIDE BINDING DETECTED BY SINGLE MOLECULE FLUORESCENCE

Using a confocal fluorescence microscope with an avalanche photodiode as detector, we studied the fluorescence of the tetramethylrhodamine labeled F1 part of the H+‐ATPase from Escherichia coli, EF1, carrying the γT106‐C mutation [Aggeler, J.A. and Capaldi, R.A. (1992) J. Biol. Chem. 267, 21355–21359] in aqueous solution upon excitation with a mode‐locked argon ion laser at 528 nm. The diffusion of the labeled EF1 through the confocal volume gives rise to photon bursts, which were analyzed with fluorescence correlation spectroscopy, resulting in a diffusion coefficient of 3.3×10−7 cm2 s−1. In the presence of nucleotides the diffusion coefficient increases by about 15%. This effect indicates a change of the shape and/or the volume of the enzyme upon binding of nucleotides, i.e. fluorescence correlation spectroscopy with single EF1 molecules allows the detection of conformational changes.

Sesquiterpene lactones as inhibitors of human neutrophil elastase

Human neutrophil elastase (HNE) is a serine protease that has been implicated in the abnormal turnover of connective tissue proteins and has been described as an important pathogenic factor in several inflammatory diseases such as rheumatoid arthritis or cystic fibrosis. Here we investigated 17 sesquiterpene lactones (SLs) for their ability to inhibit human neutrophil elastase in an in vitro assay. Podachaenin was the most active compound with an IC(50) value of 7 microM. SLs do not covalently bind to the amino acids of the catalytic triad, thus differing from other elastase inhibitors with a lactone moiety. In contrast to most other biological activities of SLs HNE inhibition is not mediated by alpha,beta-unsaturated carbonyl functions. Ligand binding calculations using the X-ray structure of HNE and the program FlexX revealed structural elements which are a prerequisite for their inhibitory activity.

Binding affinities and protein ligand complex geometries of nucleotides at the F1 part of the mitochondrial ATP synthase obtained by ligand docking calculations

F0F1 ATP synthases utilize a transmembrane electrochemical potential difference to synthesize ATP from ADP and phosphate. In this work, the binding modes of ADP, ATP and ATP analogues to the catalytic sites of the F1 part of the mitochondrial ATP synthase were investigated with ligand docking calculations. Binding geometries of ATP and ADP at the three catalytic sites agree with X‐ray crystal data; their binding free energies suggest an assignment to the ‘tight’, ‘open’ and ‘loose’ states. The rates of multi‐site hydrolysis for two fluorescent ATP derivatives were measured using a fluorescence assay. Reduced hydrolysis rates compared to ATP can be explained by the ligand docking calculations.

Synthesis and redox potentials of methylated vitamin K derivatives

We report the synthesis of derivatives of vitamin K3 as well as of vitamins K1 and K2 containing a different number of methyl groups in various positions in order to reduce their redox potentials and to change systematically their steric features. The long aliphatic chain of vitamins K1 and K2 is simulated by an undecyl chain or a methyl group, respectively. The redox potentials of the first reduction step were measured by cyclic voltammetry in DMF. These compounds are relevant for studies of the structure–function relationship of vitamin K dependent enzymes and the investigation of electron transfer reactions in photosynthetic reaction centers.

Free energy calculations on the binding of novel thiolactomycin derivatives to E. coli fatty acid synthase I.

Finding novel antibiotics to combat the rise of drug resistance in harmful bacteria is of enormous importance for human health. Computational drug design can be employed to aid synthetic chemists in the search for new potent inhibitors. In recent years, molecular dynamics based free energy calculations have emerged as a useful tool to accurately calculate receptor binding affinities of novel or modified ligands. While being significantly more demanding in computational resources than simpler docking algorithms, they can be employed to obtain reliable estimates of the effect individual functional groups have on protein-ligand complex binding constants. Beta-ketoacyl [acyl carrier protein] synthase I, KAS I, facilitates a critical chain elongation step in the fatty acid synthesis pathway. Since the bacterial type II lipid synthesis system is fundamentally different from the mammalian type I multi-enzyme complex, this enzyme represents a promising target for the design of specific antibiotics. In this work, we study the binding of several recently synthesized derivatives of the natural KAS I inhibitor thiolactomycin in detail based on atomistic modeling. From extensive thermodynamic integration calculations the effect of changing functional groups on the thiolactone scaffold was determined. Four ligand modifications were predicted to show improved binding to the E. coli enzyme, pointing the way towards the design of thiolactomycin derivatives with binding constants in the nanomolar range.

Sequence analysis reveals new membrane anchor of reaction centre-bound cytochromes possibly related to PufX

Most of the bacterial photosynthetic reaction centres known to date contain a cytochrome subunit with four covalently bound haem groups. In the case of Blastochloris viridis, this reaction centre subunit is anchored in the membrane by a lipid molecule covalently attached to the cysteine which forms the N‐terminus of the mature protein after processing by a signal peptidase. We show that posttranslational N‐terminal cleavage of the cytochrome subunit does not occur in the aerobic photosynthetic bacterium Roseobacter denitrificans. From sequence analysis of the resulting elongated N‐terminus it follows that a transmembrane helix is anchoring the reaction centre‐bound cytochrome in the membrane. Comparative sequence analysis strongly suggests that all cytochrome subunits lacking the lipid coupling cysteine share this structural feature. Comparison of the N‐terminal segment of the cytochrome subunit of Roseobacter denitrificans with the sequences of the PufX proteins from Rhodobacter sphaeroides and Rhodobacter capsulatus suggests a phylogenetic relation.

The Influence of Quinone Structure on Quinone Binding to the Q A Site in Bacterial Reaction Centers from Rhodobacter Sphaeroides

The primary reaction in the bacterial reaction center (RC) from Rhodobacter sphaeroides is the absorption of light by the primary electron donor, D, and the subsequent fast electron transfer via several intermediate acceptors, a bacteriopheophytin, ΦA, a primary ubiquinone, QA, to the secondary ubiquinone, QB. The quinone in the QB site is capable of accepting two electrons to form the hydroquinone species and exchange with the quinone pool, whereas the quinone bound at the QA site serves as a one electron gate. Compared to the QB site the quinone at the QA site is very tightly bound and does not exchange with the pool. In Rhodopseudomonas viridis and in the photosystem I of green plants a menaquinone is found as the primary electron acceptor.

The Reaction Center of Roseobacter Denitrificans: Primary Structure of the H-Subunit and Homology Model of the HLM-Complex

Roseobacter (Rs.) denitrificans belongs to the so-called aerobic photosynthetic bacteria. This type of bacteria is closely related to purple bacteria, but in contrast to these ‘true’ photosynthetic bacteria, they are not able to grow under anoxic conditions (1). Photosynthetic protein complexes and bacteriochlorophyll are only built under oxic conditions in these bacteria. Rs. denitrificans is phylogenetically close to Rhodobacter (Rb.) capsulatus a true phototrophic purple bacterium (2). The puf operon encoding the protein subunits L, M, C of the reaction center (RC) and of light harvesting complex I (LHI) from Rs. denitrificans were cloned and sequenced and show very high similarities to purple bacteria in either single sequences or the whole operon structure (3). Despite these similarities charge separation does not occur in the reaction center under anoxic conditions (4). The lack of photosynthetic activity was believed to be caused by the overreduction of the electron transfer system (1) and a higher midpoint redox potential of the primary acceptor (QA) of the RC in aerobic phototrophs compared to that of typical purple bacteria (5). For a better understanding of the charge separation in Rs. denitrificans RC the M and L subunits of RC from Rs. denitrificans were combined with the H-subunit from Rb. capsulatus in a Rb. capsulatus puf puc negative mutant (6). Charge recombination between the primary donor P3+ and Q A − was observed in the transconjugant indicating a correct assembly of L and M subunits but the transconjugant did not grow under anoxic conditions. In Rhodobacter sphaeroides it was shown that the H-subunit has an influence on electron transfer from QA to QB (7, 8). Since the H-subunit could be responsible for the differences between purple bacteria and aerobic phototrophs we cloned and sequenced the RC H-subunit from Rs. denitrificans. To get informations about the interaction between the quinones and the surrounding protein matrix a model of the HLM-complex of the RC was calculated.

Evidence for Light-Induced Cyclic Electron Transfer in Chromatophores of Roseobacter Denitrificans

Roseobacter (Rs.) denitrificans belongs to the heterogeneous group of aerobic photosynthetic bacteria which can grow and synthesize a photosynthetic apparatus only in the presence of O2 or auxiliary oxidants (1). A photosynthetic reaction center (RC), similar to those of purple non-sulphur bacteria (2), operates in cells of Rs. denitrificans catalyzing light-induced charge separation only under aerobic conditions. The RC possesses a cytochrome (cyt) c subunit which binds two high potential hemes (H1, H2) characterized by midpoint potentials (Em) equal to 290 mV and 240 mV respectively, and two low potential hemes (L1, L2) presenting a similar Em=90 mV (3). The failure of the RC photochemical activity observed in anoxic cells has been attributed to a relatively high midpoint potential and oven-eduction of the primary acceptor QA in the RC (2). Although participation of a bc1 complex in cyclic electron transfer has been suggested by studies in intact cells (3), the reaction pathways sustaining light-induced electron transfer are poorly understood. In this work flash-induced absorption changes arising from b-type and c-type hemes were studied in Rs. denitrificans chromatophores under controlled redox conditions. The data support a Q-cycle mechanism.