Open Access

Show Abstract ▼

Abstract: Tripartite motif-containing protein 21 (TRIM21), and particularly its PRY-SPRY protein interaction domain, plays a critical role in the immune response by recognizing intracellular antibodies targeting them for degradation. In this study, we performed a crystallographic fragment screening (CFS) campaign to identify potential small molecule binders targeting the PRY-SPRY domain of TRIM21. Our screen identified a total of 109 fragments binding to TRIM21 that were distributed across five distinct binding sites. These fragments have been designed to facilitate straightforward follow-up chemistry, making them ideal starting points for further chemical optimization. A subsequent fragment merging approach demonstrated improved activity. To enable functional validation of compounds with full length human TRIM21, we established a NanoBRET assay suitable for measuring target engagement to the main Fc binding site in life cells. The high-resolution structural data and observed binding modes across the different sites highlight the versatility of the PRY-SPRY domain as a target for small-molecule intervention. The presented data provide a solid foundation for structure-guided ligand design, enabling the rational design of more potent and selective compounds, with the goal to develop bivalent molecules such as Proteolysis Targeting Chimeras (PROTACs).

Show Abstract ▼

Abstract: Bradyrhizobium japonicum USDA110 is a widely used microorganism in the formulation of bioinoculants for soybean crops, harboring a copper-containing nitrite reductase with low enzymatic activity. The activity of BjNirK at pH 6.5 was higher compared to that at pH 8.0, regardless of the presence of either physiological or artificial electron donors. Thermal shift assays reveal that the enzyme is more stable at pH 6.5 than at pH 8.0. X-ray structural data reveals that the funnel for substrate entry shows a wider cavity when compared to other class I NirK structures. Furthermore, the presence of an additional channel for proton provision is observed, in addition to the primary and secondary proton channels. The T2Cu active site can accommodate one or two water molecules, resulting in a tetra- or pentacoordinated metal site, respectively. The structural data correlates well with both optical visible and resonance Raman spectroscopies, denoting a strong blue character of the T1Cu site in both solid and solution states. Furthermore, EPR-monitored redox titration reveals that the catalytic rate is not constrained by T1Cu−T2Cu intraprotein electron transfer reaction at either pH 6.5 or pH 8.0. Additionally, bioinformatics studies indicate that the interaction between the enzyme and the electron donor is not pH dependent. These two observations suggest that the low activity of BjNirK is not caused by inefficient donor−enzyme interaction or impaired electron transfer. The present results suggest that the structural architecture and enzyme properties in rhizobia are designed to ensure low activity, a trait that is particularly advantageous for symbiosis.

Show Abstract ▼

Abstract: The discovery of new drugs is essential for human well-being, as effective treatments are crucial for combating diseases, addressing emerging health threats, and improving overall quality of life. The continuous development of novel therapeutics is particularly important in response to evolving pathogens, drug resistance, and unmet medical needs. However, drug discovery is a complex and resource-intensive process, given the intricacy of biological systems and the vast chemical space of potential drug candidates. Traditional wet-lab experiments are costly and time-consuming, making the search for novel drugs akin to finding a needle in a haystack. In this thesis, we investigate how machine learning systems can support early-stage drug discovery in scenarios where experimental data is scarce. First, in a practical study addressing a real-world low-data challenge - namely, the search for potential drug candidates during the COVID-19 pandemic - we apply and evaluate a zero-shot learning method. Furthermore, the thesis introduces a novel few-shot learning architecture, MHNfs, which enhances molecular representations through a memory-based mechanism that we refer to as context enrichment, where contextual information is retrieved from an external memory. The work also connects recent few-shot learning approaches in drug discovery with the concept of in-context learning, originally introduced for large language models. Finally, to support real-world applications, the thesis presents an interactive interface that enables chemists and drug designers to easily provide prompts to and interact with the MHNfs model, facilitating the integration of few-shot learning into their workflows. Overall, the thesis contributes to the advancement of few-shot learning methods for drug discovery and their practical applications in real-world scenarios.

Open Access

Show Abstract ▼

Abstract: Crystallographic fragment screening is a powerful methodology that enables the identification of low molecular weight ligands and has shown great promises in drug discovery. In this work we report the results of a fragment screening carried out in an effort to further map the cavities of trypanothione reductase from Trypanosoma brucei (TbTR), a critical target for drug design against human African trypanosomiases (HAT), for which efficient and non-toxic trypanocidal drugs are lacking. Moreover, the conservation of trypanothione reductase among trypanosomatids, including Leishmania, could facilitate the design of a wide-spectrum drug against many parasitic diseases. At the XCHEM facility (Diamond Light Sources, United Kingdom) we performed the soaking of TbTR monoclinic crystals with fragments from DSIpoised and EubOPEN DSIp libraries and we identified eight new hits binding to different cavities of TR including the trypanothione and the NADPH binding cavities. These fragments exhibited affinities ranging from submillimolar to millimolar, as determined by surface plasmon resonance (SPR). While the newly identified fragments did not significantly alter TbTR’s enzymatic activity—consistent with the nature of low-affinity ligands—they provide valuable insights into key interactions of fragments with TR and, together with prior fragment screening campaigns, pave the way towards follow-up chemical optimization into lead compounds.

Open Access

Show Abstract ▼

Abstract: SARS-CoV-2 Mpro is a cysteine protease, that acts as a symmetrical dimer and displays positive cooperativity for substrate turnover. A series of potent reversible covalent peptidomimetic aldehydes and nitriles was designed as Mpro inhibitors. To better understand the observed structure activity relationships (SAR) binding potency and mechanism was examined by enzyme activity assay, surface plasmon resonance, X-ray crystallography, matrix-assisted laser desorption electrospray ionization and NMR. Potent aldehydes bind Mpro cooperativity but bind covalently to only one subunit of the dimer. The analogous nitriles do not bind cooperatively, and the degree of covalent binding observed varied depending on the assay method employed. The NMR studies support that potent inhibition of Mpro by the nitriles does not require covalent binding. The data highlight the caveats in using orthogonal assays to confirm compound mechanism, particularly in cooperative systems.