Scientific data are typically published in a written form (journal articles, theses). Author of such text should not only present the data correctly, but also follow standard rules of (usually English) grammar, style, and typography. The latter topic is briefly reviewed by L. Zidek in his presentation. The presentation covers necessary background information and terminology, and recommendation how to typeset a scientific text (with equations and tables) and how to prepare figures.

Tyrosine hydroxylase (TH) catalyzes hydroxylation of L-DOPA and it is a rate-limiting enzyme in biosynthesis of important neurotransmitters and hormones: dopamine, noradrenaline and adrenaline. Its activity is regulated by phosphorylation of its regulatory domain and by the interaction with 14-3-3 protein. So far the full length TH structure was not solved and the activation mechanism stays unclear. The aim of my diploma thesis is to solve the structure of human TH in apo state and in complex with 14-3-3ζ protein using single particle cryo‑electron microscopy. My research work covers protein purification, grid preparation, data collection and 3D reconstruction.

Lectins are ubiquitous sugar binding proteins serving a wide range of purposes in both physiological and pathological processes across animals, plants, bacteria, fungi, protozoa and viruses, e.g control of synthesis and subcellular delivery of glycoproteins, cancer, cell-cell recognition, etc¹. Importantly, they frequently mediate pathogen adhesion to host cells, such lectins can be targeted by antagonists to abolish the infection initiation².
Photorhabdus asymbiotica is a gram-negative bioluminescent entomopathogenic bacterium. As the only species in Photorhabdus genus, P. asymbiotica is also capable of infecting human and is classified as an emerging human pathogen³.
My diploma work focuses on PHL2 and PHL3 lectins from P. asymbiotica, both sequence homologs of PLL lectin from P. luminescens also studied in our group⁴. Various biophysical techniques were used to characterize recombinant PHL2 and PHL3 lectins: structural insights are provided by CD spectroscopy, analytical ultracentrifugation (AUC) and X-ray crystallography, protein stability by differential scanning fluorimetry (DSF) and interaction with various ligands by isothermal titration calorimetry (ITC), surface plasmon resonance (SPR) and glycan array. Together these methods enable understanding of protein function and shed light on their involvement in the infection process.

The talk focuses on computer modeling of the tear film lipid layer (TFLL), a relatively thin lipid-rich region covering the human eye. The main role of TFLL is to reduce surface tension of the tear film. Furthermore, it helps with the tear film re-spreading after blinks, provides a smooth optical surface, and prevents water evaporation. TFLL deficiencies lead to evaporative dry eye syndrome, one of the commonly reported eye ailments. Composition and structure of the TFLL are still under debate. In the talk, our in silico TFLL model which is able to capture key molecular-level characteristics of the tear lipid film, will be discussed. By employing this model in coarse grain MD simulations, we investigated overall structure of TFLL including the role of individual lipid classes as well as interactions of proteins with the lipid film.

In the talk, I will run you through a couple of examples from "exotic" languages like Tabasaran (the language with most cases on the planet!), Malayalam and others to show you how location and motion in space is coded in natural languages. What I want to show is that once properly understood, the surface diversity of such languages is only apparent: there is a fairly simple and rigid combinatorial system underlying all such languages. Linguists call this shared combinatorial system "The Language of Thought" or "Universal Grammar." The idea has been around for a while, the new thing I will focus on is the development of the field into a more and more fine-grained understanding of this component of the human mind. Put simply, what we are discovering is that the basic building blocks of language are very small, much smaller than traditionally thought.

Nanodisc is a patch of phospholipid membrane stabilized by scaffold proteins such as Apo lipoprotein. It is a biotechnological tool providing a natural lipid environment for isolation, purification, and structure determination of transmembrane proteins. However, the nanodisc is not suitable for all proteins as the length of current scaffold protein limits its size. We present a new nanodisc with easily adjustable size, where the scaffold protein is based on repeating helical building blocks. These helical constructs were designed to spontaneously orient perpendicular to the membrane plane and to stack together forming a protein fence around the membrane disc. The design is based on our coarse-grained molecular dynamics simulation, where the aromatic residues on specific positions together with the width and length of the hydrophobic patch on the helices were found to be crucial for the orientation. The de novo designed protein provides a novel scaffold template for formation of membrane nanodiscs with easily adjustable size and demonstrates the strategy for design of new proteins for membrane edges.

Transcription is a process leading to production of RNA. RNA polymerase II (RNAPII) is the enzyme in charge of both, coding (mRNA) and non-coding (snRNA, snoRNA) gene transcription. However, thanks to intrinsic bidirectionality of most promoters and poor regulation of transcription initiation, RNAPII transcribes also wide range of anti-sense, sense or ORF-shifted transcripts, which are rapidly recognised and such “pervasive” transcription is aborted. The main machinery controlling transcription of non-coding RNA in eukaryotic cell is the Nrd1–Nab3–Sen1 (NNS). During the transcription termination of non-coding RNA (ncRNA) genes, the NNS complex is recruited to the transcribing RNA polymerase II through the recognition of specific motifs in the nascent RNA by Nrd1 and Nab3, and the interaction of Nrd1 with the C-terminal domain of RNA polymerase II. The assembly and dynamics of Nab3 and Nrd1 functional domains and its impact on RNA binding remains poorly understood. The main goal of the thesis is the study of Nrd1 and Nab3 heterodimerization and effect of the dimerization on interdomain dynamics. Experimental work included design and determination of dimerization domain borders using bioinformatics, optimization of expression and purification of individual domains, developing the procedure for the formation of heterodimer, biophysical characterization of studied domains using size exclusion chromatography, CD and SAXS.

Delta is the accessory subunit of bacterial RNA polymerase, present only in Gram-positive bacteria. Although the delta subunit is dispensable part of RNAP and the cells lacking the delta are viable in pure cultures, this protein is required for rapid changes in gene expression and competitive fitness of the cells in the natural environment. The delta subunit affects the sensitivity of RNAP to initiating NTPs, thus enhances transcription specificity. The detailed molecular mechanism of its activity is still unknown. The delta subunit contains disordered C-terminal domain (CTD) with multiple repetitions of negatively charged residues. The CTD also comprise the conserved Lysine stretch, generating positive charge. Previous investigation of delta subunit revealed transient electrostatic interactions between positively and negatively charged residues in the CTD. We investigated the influence of the electrostatic interactions on the global shape and structural behavior of the delta subunit using various NMR experiment, such as measurements of residual dipolar coupling, paramagnetic resonance enhancement and relaxations. The results revealed that electrostatic interactions influence folding of the delta subunit and govern the motional properties of the CTD.

Antimicrobial peptides (AMPs) belong to a broad group of membrane-active compounds. Despite being diverse in their structure and origin, they share common characteristics, such as amphiphilicity or positive net charge. These features are key aspects for the pathogen recognition. AMPs can disrupt membrane of foreign organisms through the formation of membrane pores. AMPs have thus become attractive drug targets in development of new antibiotics. However, the optimal conditions, leading to the formation of membrane pores, still remain obscure. We are particularly interested in AMPs with α-helical secondary structure and the effect of peptide kink. In our coarse-grained model, each α-helical region of a peptide is represented by a single particle with attractive (i.e. hydrophobic) patch.^[1] In total, we have systematically prepared 144 peptide models to describe different AMPs. We performed Monte Carlo simulations with Metropolis scheme. Wang-Landau method was used to calculate the free energy profiles associated with pore formation under various conditions. We evaluate the effect of kink flexibility and angle, properties of peptide termini, hydrophobic content, and concentration (defined as peptide-to-lipid ratio). Using this systematic approach, we reveal the contribution of each aspect to pore formation. Primarily, we demonstrate that the peptide kink decreases the free energy required for pore formation. Furthermore, various P/L ratios, peptide conformations and different proportion of hydrophobic content determine geometrical arrangement of peptides within a pore, showing that peptides assume such an orientation, which effectively stabilizes membrane curvature.

Finally we present several plausible pore models that exert similar characteristics as experimentally observed pore structures. These findings can be utilized as templates for novel therapeutic candidates.
 
[1] Vácha R, Frenkel D. Biophysical Journal. 2011;101(6):1432-1439.

Oligomers of the β-amyloid (Aβ) peptide are thought to be implicated in Alzheimer’s disease (AD). Membranes of neurons may mediate the oligomerisation of Aβ present in brain. Using the single-molecule sensitivity of fluorescence, our work addresses the oligomerisation of Aβ monomers on lipid bilayers containing essential components of the neuronal plasma membrane. In this talk, using Fluorescence Correlation Spectroscopy (z-scan FCS and cross-correlation FCS) we will find that sphingomyelin triggers the oligomerisation of Aβ and that physiological levels of GM1, organized in nanodomains, prevent oligomerization of Aβ thus counteracting the effect of sphingomyelin. Our results demonstrate a preventive role of GM1 in the oligomerisation of Aβ suggesting that decreasing levels of GM1 in the brain, e.g. due to aging, could lead to reduced protection against Aβ’s oligomerisation. In addition, we show how the existence of nanoscopic heterogeneities (radius 8-26 nm) in microscopically homogeneous membranes, which are unresolvable by super-resolution microscopy, is possible through a combination of Monte Carlo Simulations and FLIM-FRET experiments.

Partial atomic charges are real numbers describing the distribution of electron density in a molecule, thus providing clues to the chemical behaviour of molecules. The concept of charges began to be used in organic chemistry and physical chemistry. Afterwards, partial atomic charges were adopted by computational chemistry and molecular modelling (e.g., molecular dynamics, docking, conformational searches, binding site predictions). Recently, the charges started to be used also in chemoinformatics and structural bioinformatics.
To provide the life science community with an easy access and fast visualization of partial atomic charges, we have developed the ChargeViewer software tool. It enables a user to read charges from all available charge data formats and to visualize them via all standard charge visualization models. ChargeViewer provides high-quality visualization and can quickly show even charges on surfaces of large biomacromolecules. We demonstrate its applicability on three case studies. The first study is focused on the relation between charge distribution and acid dissociation constant. The second study demonstrates a similarity analysis of sugar binding sites in lectins. The third study examines the activation mechanism of pro-apoptotic protein BAX.

The acid dissociation constant (pKa), and the partition coefficient (logP) are important physico-chemical properties whose values were applied in several chemical, biochemical, pharmaceutical or environmental studies. A very promising approach for their prediction is the utilization of QSPR (Quantitative Structure-Property Relationship) models, where partial atomic charges are applied as descriptors. Partial atomic charges are real numbers which describe the distribution of electron density in a molecule. In spite of the fact they can not be measured experimentally, several different quantum mechanical and empirical approaches to calculate these charges were created. The charge descriptors utilized in QSPR modeling are calculated from partial atomic charges of one molecule and results in sets of numbers which can describe a particular aspect of the molecule. The main goal of this master's thesis is the preparation of QSPR models for prediction of pKa and logP properties, that will be applicable to the large-scale molecular sets. An essential step is the preparation of a sufficiently large database for the training of QSPR models. We introduce two newly created databases of pKa and logP values, created by joining three available databases HSDB, PHYSPROP, and QSAR Toolbox. Another important aspect of physico-chemical properties prediction are descriptors used for QSPR modeling. Therefore we introduce an ChrgDescCalcII.py, the tool for charge descriptors calculation. Tool has implemented 148 charge descriptors in total. Afterwards, several types of charges were used for design of QSPR models (quantum mechanical and empirical) and two statistical approaches, namely multiple linear regression and random forest. The best created QSPR model for pKa prediction achieved reproductive statistics of R² = 0.942. The best created QSPR model for logP prediction achieved reproductive statistics of R² = 0.913.