Spännande läkemedelsforskning

Abstract

Microscopy is an integral technique in biology to study the fundamental components of life visually. Digital microscopy and automation have enabled biologists to conduct faster and larger-scale experiments with a sharp increase in the data generated. Microscopy images contain rich but sparse information, as typically, only small regions in the images are relevant for further study. Image analysis is a crucial tool for biologists in the objective interpretation and extraction of quantitative measurements from microscopy data. Recently, deep learning techniques have shown superior performance in various image analysis tasks. The models learn feature representations from the data by optimizing for a task. However, the techniques require a significant amount of annotated data to perform well. Domain experts are required to annotate microscopy data, making it expensive and time-consuming. The models offer no insight into their prediction, and the learned features are not directly interpretable. This poses challenges to the reliable utilization of the technique in high-trust applications such as drug discovery or disease detection. High data variability in microscopy and poor generalization performance of deep learning models further increase the difficulty in general usage of the technique. 

The work in this thesis presents frameworks and methods to solve the practical challenges of applying deep learning in microscopy. The application-specific evaluation approaches were presented to validate the approaches, aiming to increase trust in the system. The major contributions of this work are as follows. Papers I and III present human-in-the-loop frameworks for quick adaption of deep learning to new data and for improving models' performance based on human input in visual explanations provided by the model, respectively. Paper II proposes a template-matching approach to improve user interactions in the framework proposed in Paper I. Papers III and IV present architectural modifications in the deep learning models proposed for better visual explanation and image-to-image translation, respectively. Papers IV and V present biologically relevant evaluations of approaches, i.e., analysis of the deep learning models in relation to the biological task.

This thesis is aimed towards better utilization and adaptation of the DL methods and techniques to the microscopy data. We show that the annotation burden for the user can be significantly reduced by intuitive annotation frameworks and using contemporary deep-learning paradigms. We further propose architectural modifications in the models to adapt to the requirements and demonstrate the utility of application-specific analysis in microscopy.

Abstract

The thioether motif is found in numerous pharmaceuticals. The simplest form of this motif, the methyl-thioether is similarly found in many biologically active compounds and has exhibited many advantageous properties. The installation of the thiomethyl moiety can be troublesome due to the toxicity and malodor of small sulfur reagents. To avoid these issues, many alternative sources and methods have been developed, although these often require metal catalysis, or operate via electrophilic species as the sulfur source.

The first section of this thesis covers the discovery and development of BF₃SMe₂ as a Lewis acid and non-malodorous source for nucleophilic installation of the thiomethyl moiety. In paper I, this reagent is leveraged for the Lewis acid activation and the nucleophilic thiomethylation of electron-deficient haloarenes. The reagent was also found to be a selective reductant for nitropyridines and could in some instances perform concomitant thiomethylation via C-H substitution. The nucleophilic installation of the thiomethyl moiety was continued in paper II, where BF₃SMe₂ was developed for the conversion of aromatic aldehydes into methyl-dithioacetals. The dithioacetal is an important protecting group, and a useful intermediate, and since there are only a handful of reported strategies for the synthesis of the methyl-analog, this method represents an important addition to existing methods. In addition, BF₃SMe₂ was able to promote Friedel-Crafts reactions between aldehydes and electron-rich arenes, resulting in unsymmetrical, thiomethylated diarylmethanes. 

In the final part of this section, BF₃SMe₂ was used for the activation of trifluoromethylarenes for defluorination and sulfur incorporation, resulting in the methyl-dithioester moiety. This functional group is a useful intermediate, but current methodology for its synthesis suffers from several drawbacks, including multistep reactions. The method developed in paper III is a convenient one-step approach to reach the methyl-dithioester, and was also expanded to one-pot synthesis of thioamides and different heterocyclic systems starting from the trifluoromethyl moiety.

In the second section of this thesis, an in situ method for diazomethane release and consumption was investigated and applied. Diazomethane is a useful reagent with a unique reactivity profile, including mild and selective O-methylation of carboxylic acids. The reactivity of this gaseous reagent however comes with hazards, such as high toxicity. Therefore, earlier efforts have been made for in situ generation of diazomethane, although with limitations such as specific solvent requirements or alkaline conditions. The method developed in paper IV however is an improvement upon these with base-free conditions, and wide solvent compatibility, and was successfully applied to the methylation of carboxylic acids, solvent-mediated deuterium labeling without any a priori deuterium incorporation, and synthesis of diazoketones.

Abstract

Capillary electrophoresis with mass spectrometric (CE–MS) detection offers a separation method without equal in terms of flexibility, utility, and cost efficiency. Here we demonstrate precisely this through the application of several laboratory-built CE–MS instruments for the separation of brain metabolites in non-primates, enantioselective separations of synthetic anesthetic metabolites in fractionated pony urine, application in structural proteomics workflows, and identification of exogenous alkaloid biotransformationproducts in human cerebrospinal fluid (CSF).

We outline a method for quickly and affordably etching austenitic steel tubing, which is widely used in electrospray sources for CE–MS. The stainless steel tapered tip emitters provide robust electrospray with low sheath liquid flow rates and can be easily fabricated in-house, offering flexibility and cost-efficiency when commercial options areunavailable. We contribute a CE–MS method for enantiomer separation, specifically targeting 6-hydroxynorketamine (HNK). By introducing chiral selectors into the separation capillary, the method enables efficient enantiomer separation and offers a newtool to assist with research on HNK as a cure for depression.

We explore the feasibility of cold CE–MS in hydrogen deuterium exchange workflows. The utilization of a lab-designed Peltier-cooled CE device achieves deuterium back exchange rates on par with commercial liquid chromatography-based platforms, offering new possibilities for studying protein structures and interactions.

We also demonstrate the wide ranging versatility of CE–MS with contributions to the identification of specific tobacco related metabolites in CSF samples during the development of a high throughput mass spectrometry diagnostic tool for Parkinson’sDisease.

This thesis showcases the versatility and value of CE–MS in various applications, a true blessing for analytical chemistry.