Application of Computational Methods to the Structural and Functional Properties of Flexible Chiral Molecules

Language
en
Document Type
Doctoral Thesis
Issue Date
2016-05-31
Issue Year
2016
Authors
Brkljaca, Zlatko
Editor
Abstract

Flexibility and chirality of molecules are two cornerstones of modern chemistry, underlying vast array of biochemical and biological processes that take place in living organisms. The former concept and its role in this respect has been investigated and documented for over 70 years, including its critical role in the interactions occuring between enzymes and their substrates and, more generally, between receptors and ligands. Flexibility thus represents one of the mainstays of a generalized catalytic mechanism. While the concept of chirality has been introduced more than a century ago, certain manifestations of this phenomenon have been properly investigated for a relatively short period of time and, despite recent experimental and theoretical efforts, the complete understanding of this phenomenon is yet to be accomplished. Interestingly, although it is precisely the interplay between the two that plays a crucial role in the fields of biochemistry and molecular medicine/pharmacology, e.g. being necessary for the development of conceptually novel systems for targeted drug delivery, it is also significantly more scarcely investigated and understood than either of the two underlying concepts.

The aim of this thesis is to investigate the interaction between flexibility and chirality using state-of-the-art computational methods, ranging from highly accurate quantum mechanical treatments to more approximate classical methods. We have thereby investigated the interplay of the two concepts in two different scenarios, dealing with the behavior of flexible and chiral molecular species in solution (in the bulk of the solvent) and on the inorganic surface, respectively.

In the first part of this thesis we focused on the flexible, chiral molecules in solution, which we decided to inspect using the most powerful manifestation of chirality, namely using circular dichroism (CD) spectroscopy. CD is one of the key experimental methods employed in the structural characterization of optically active chiral molecules and, as such, it is widely used in studies of biologically important systems, such as peptides, proteins and DNA molecules. However, extracting useful information from experimentally obtained CD spectra can be a complex task, especially in the case of highly flexible molecules. Thus, to obtain a detailed understanding of the experiments, theoretical methods and resources need to be employed. In this respect, we developed a general methodology for calculating the CD spectra of flexible molecules by combining two “worlds”, namely classical and quantum mechanics. Our method is based on the generation of a converged conformational phase space, obtained from advanced classical molecular dynamics simulations, namely replica exchange molecular dynamics, which is followed by finding a set of structures representing the entire phase space using clustering analysis. In agreement with previous studies, we have observed strong solvent effects on the spectra. These were taken into account by calculating an average field consisting of point charges originating from a number of solvent configurations around each peptide conformation in the subset. Subsequently, the CD spectrum of each solvated conformation was obtained using the quantum mechanics/molecular mechanics (QM/MM) method incorporating a time-dependent density functional theory (TD-DFT) calculation. The average CD spectra were obtained by taking the mean of all weighted single CD spectra, where the weighting factor of an individual spectrum is given by the cluster-generated population fraction represented by the respective conformation. Qualitatively, our procedure can be considered as the sequential averaging over the solvent (for each representative conformation), the sidechains (inside the backbone-based clusters), and the backbone (combining the principal clusters). The verification of our methodology came in terms of two separate case studies, which are elaborated in more details in the following paragraphs. 

In our first case study we applied the developed methodology to calculate the CD spectra of two novel Rhodomyrtal compounds which possess potent antibacterial activity. With the knowledge of the relative configurations, known from NMR techniques, the task at hand was to determine the absolute configuration of the two compounds.  We were thereby able to sucessfully reproduce the experimental CD spectra using our strategy and utilizing TD-DFT method using B3LYP /6-311G(d) level of theory, obtaining an excellent agreement between experimentally measured and calculated CD spectra. More precisely, the three most dominant spectral features exhibited experimentally by these species are both qulitatively and quantitatively reproduced by the theoretically obtained CD spectra. In tur, this enabled us to unambigously assign the absolute configurations of both Rhodomyrtal compounds.

We followed with our second case study, in which we used the developed methodology in an attempt to calculate CD spectra of the two highly flexible opioid pentapeptides, which are found to exhibit a pronounced anti-tumor activity, namely Met- and Ada-enkephalin (Tyr-Gly-Gly-Phe-Met and Tyr-Aaa-Gly-Phe-Met, respectively, where Aaa denotes an unnatural adamantyl containing amino acid). We therefore modeled both the zwitterionic and neutral forms of the enkephalins and additionally both the R- and the S-epimers of Ada-enkephalin, which was necessary due to the fact that zwitterionic and neutral forms of the peptides coexist in trifluoroethanol, also taking into account the fact that the available experiments were conducted on an epimeric mixture of Ada-enkephalin. Thus, to make an appropriate comparison with experiment, we have produced composite spectra that account for the appropriate contributions of the zwitterionic and neutral forms of the peptides, as well as the expected epimeric ratio in the case of Ada-enkephalin. Despite the complexity of the task, qualitative agreement with experiment has been obtained below 230 nm and the main features of the measured spectra have been reproduced by our calculations. However, the simulated spectra show a persistent minimum between about 230-240 nm that is not present in the experimental measurements and which we initially assumed to arise from the poor treatment of a particular excited state by TD-DFT. Motivated both by this finding and by recent studies that have pointed to problems with modeling charge transfer excitations, we benchmarked the performance of the selected global hybrid functionals (GHFs), namely B3LYP and PBE0 functionals, long-range corrected functionals (LCFs), more precisely CAM-B3LYP and ωB97X-D functionals, and one hybrid meta functional (HMF), namely M06-2X, against high level ab initio RICC2 calculations for selected peptide structures.  Furthermore, we compared the performance of the functionals with the experimentally available data. Our results show that long-range corrected functionals correlate relatively well with RICC2 calculations, as does the meta-hybrid M06-2X, while the global hybrid functionals exhibit the aforementioned charge-transfer artifacts. Moreover, PBE0 and even more so M06-2X and B3LYP produce spectra in better agreement with the experimental data. We have clarified this apparent discrepancy by finding that the surplus charge-transfer excitations, exhibited by B3LYP and PBE0, seem to have a negligible contribution to the final spectra, once appropriate structural averaging is performed. This implies that the averaging procedure substantially reduces the effects of the problems exhibited by the GHFs in the CD spectra of the individual model structures. In this respect, our study strongly suggests that TD-DFT is likely to be preferred compared to the stronger RICC2, not only because of its better agreement with experiment for certain transition classes but also because it can be conveniently used, even for relatively large molecules, to evaluate the substantial number of conformations required to achieve converged CD spectra for flexible molecules. 

In the second part of the thesis we concentrated on the behavior of flexible chiral organic molecules at the water – biomineral interface, in an attempt to uncover the roles of flexibility and chirality in biomineralization and biomineralization-inspired drug design. We thus decided to explore the interactions occurring within a potential biomineral – biomolecule composite using advanced molecular dynamics simulations, which enabled us to obtain atomistic details at the interface between biomolecules and mineral surfaces. More precisely, using advanced simulation techniques we investigated the adsorption behavior of two epimeric peptides, namely R- and S-Sal (N-Sal-Gly-S-Asp-R-Asp-S-Asp and N-Sal-Gly-S-Asp-S-Asp-S-Asp respectively, where N-Sal denotes the N-terminal residue which is a salicylic acid derivative), on both the stabile (104) and growing (001) surfaces of calcite. The reasoning behind this particular biomolecule – biomineral combination is twofold; on one hand, it represents a prototypical example of the biodegradable drug delivery system, with calcite serving as the matrix/carrier and salicylic acid playing the role of an archetypical drug, while the reminder of the peptide acts as the linker through which the drug attaches to the inorganic carrier. On the other hand, the studied epimers were experimentally shown to inhibit and modify the growth of calcite, showcasing that they play a significant role in the biomineralization of calcite.


Our research suggests that neither epimer binds to the stabile surface of calcite, while both the R- and S-epimers adsorb very strongly to the calcite (001) surface. R-Sal was found to bind more favorably, with the difference in the calculated adsorption free energies between the two forms amounting to approximately 85 kJ·mol−1. In both cases the epimers adsorb to the positively charged half of the growing (001) calcite surface through either three or four available negatively charged carboxyl groups. We explained the observed difference in free energies of adsorption between two epimers by noting that the R-epimer has a stronger electrostatic interaction with the (001) calcite surface, exhibiting a predominant adsorption mode in which all four carboxyl groups interact with Ca2+ rich surface. More precisely, approximately 15% of S-Sal structures are found to be adsorbed via three carboxyl groups, thus lowering the enthalpic contribution to the free energy. This finding agrees very well with the experimental measurements, where it was found that R-epimer inhibits/modifies the growth of the calcite mineral more strongly compared to its epimeric counterpart. Our work is the first theoretical study to shed light on the role of the chirality and flexibility on biomineralization and biomineralization-inspired drug design. Our results suggest that chirality can be used, on one hand, as a fine-tuning tool via which one can monitor and modify the growth of the biomineral, while, on the other hand, it may also serve to change and tweak the drug delivery rates by varying the strength of the linker – biomineral interactions, in turn changing the pharmacodynamical response and therapeutic efficacy of the targeted drug delivery systems.

The results of this thesis emphasize the power of state-of-the-art theoretical methods, which we successefully used to illuminate the roles of chirality and flexibility, both for the case of the flexible, chiral molecules in the solution and at inorganic/mineral interfaces. We thereby elucidated the interaction between these two crucial concepts by establishing the bridge between the molecular structure and spectra of highly flexible species for the first time. The interplay between the two has also been investigated in the contexts of biomineralization and biomineralization-inspired drug design. In that context, we were thereby able to clarify the role of flexibility and chirality in the interactions between biomolecules and biomineral surfaces by calculating free energy profiles of adsorption using advanced classical molecular dynamics simulations. Importantly, we find that the interplay between the theory and the experiment is not only helpful, but rather necessary when trying to properly interpret the experimental measurements, which, in turn, enable us to validate and further develop existing theories and models.
DOI
Faculties & Collections
Zugehörige ORCIDs