Browsing by study line "Genetics and genomics"

The aim for this project is to set up a high-content imaging pipeline for phenotypic analysis of single cells in peripheral blood mononuclear cell (PBMC) samples from healthy blood donors. The blood donors selected for the optimization experiments are known to carry specific allele variants of interest, based on an earlier FinnGen study. The main question is whether these genetic differences result in phenotypic changes in the PBMCs that can be identified by microscopic imaging and AI-guided image analysis. In this Pro Gradu work, I have optimized the pipeline of PBMC sample handling, immunostaining, and phenotypic imaging. PBMCs were gathered from healthy donors at the Blood Service Biobank. The frozen PBMC samples were thawed, and cells were plated on 384-well plates prior to immediate fixation with paraformaldehyde. The cells were then stained with fluorescent cell markers based on the Cell Painting assay (Bray et.al. 2016), followed by wide-field and confocal imaging with Opera Phenix high-content confocal microscope (FIMM High Content Imaging and Analysis unit). Novel deep learning methods are now being developed (Pitkänen group) to automatically learn phenotypes from the collected imaging data and associate them to the donor’s genotypes. We also used in-house tools for cell segmentation and further analysis as well as quality control (Paavolainen group). Primary results based on the features extracted from acquired images showed promising cell type - and donor -type specific clustering.

Distal myopathies are a group of rare progressive genetic muscle disorders that are extremely varied both genetically and clinically. Typical symptoms include weakness and atrophy limited to the skeletal muscles of distal extremities in hands and legs. The age of onset ranges from early childhood to late adulthood depending on the disease. Currently around 30 genes have been associated with distal myopathies, most of them causing a dominant disease. The objective of the thesis was to identify the disease-causing variant in a family affected by autosomal dominant distal myopathy with early adulthood onset. Affected family members expressed weakness and atrophy in muscles of both hands and legs. To narrow down the chromosomal location of the disease-causing variant, linkage analysis was conducted with genome-wide single nucleotide polymorphism data of family members. Because of the progressive nature of the disease and uncertain disease status of one family member, linkage analysis had to be repeated a few different times with different settings. Both disease statuses and pedigree size were altered to account for the possibility of presymptomatic carriers or incomplete penetrance. Analyses with different parameters led to discovery of multiple possible co-segregating regions. Rare co-segregating small-scale and structural variants as well as repeat expansions in these regions were examined from next-generation sequencing data with multiple bioinformatic detection tools. The segregation of possible candidate variants was validated with Sanger sequencing and PCR. Ultimately, no likely rare co-segregating variant of any type of genetic variation with a likelihood to cause a disease such as distal myopathy was identified by any detection method used. Lack of potential disease-causing variant could be due to incomplete penetrance of the variant or if it was in non-coding regions, such as a deep intronic splicing variant in a gene currently not known to be connected to muscles.

The composition and dynamics of the early life gut microbiota plays a major role in establishing neonatal immunity and is suggested to have multiple impacts on the child’s long-term health. Meanwhile, the composition of the infant gut microbiome has been shown to be affected by the birth mode, infant health and diet. However, the characterization of the infant gut microbiome and its impact on the host’s health is still challenging as the contribution and importance of multiple co-factors on the early microbiome during infant growth is still poorly understood and characterized. The Health and Early-life microbiota (HELMi) is a cohort of more than 1000 healthy Finnish infants currently followed from birth to 4-5 years old. By now, the HELMi dataset comprises more than 400 whole genome shotgun metagenomes obtained from stool samples from 80 infants and parents, but also an in-depth characterization of the families’ lifestyle, environment, health and nutrition, allowing for a precise and cutting-edge characterization of the early gut microbiota. Based on the datasets from the HELMi, this project used Metaphlan3, Kraken and Braken to determine the best computational approach for the taxonomic profiling of the metagenomic reads. Then a PERMANOVA test was performed to evaluate and determine the factors significantly associated with the compositional microbiota variation within the infant gut metagenomes. This study first identified technical factors introducing bias in taxonomic profiling (e.g., DNA extraction batch), which served as confounders in the analysis of environmental and host variables. The investigation of these biological factors indicates that pre-natal and peri-natal variables such as the mode of delivery significantly impact the infant gut microbiota, while we did not identify any significant impact of breastfeeding habits and medication exposures in this study.

The ends of eukaryotic chromosomes are formed by a special heterochromatic structure, the telomere, which is essential to guarantee chromosome stability. Telomeres protect chromosomic ends from DNA degradation, repair, and recombination events. However, they are difficult to replicate due to their repetitive and heterochromatic nature, which hinder DNA replication fork progression. In yeast, Mph1 helicase promotes replication fork regression, cross-over suppression during homologous recombination (HR), and telomere maintenance. Moreover, Mte1 is a D-loop binding protein involved in response to DNA damage and maintenance of telomere length, which interacts with Mph1, thereby stimulating its regression capacity as a helicase and fork. Thus, the Mte1-Mph1 complex is recruited to stressed telomeres. Mte1 also shares a domain of unknown function, DUF2439, with Rad51 and Rdh54. Additionally, Esc2 protein is involved in the regulation of DNA damage through template switch (TS) recombination, preventing HR events caused by Mph1. This thesis aimed to uncover the potential roles and interactions of proteins involved in telomere maintenance, such as Mph1, Mte1, Esc2 and Rdh54, for which two main assays were conducted: (1) Telomere Stability assay, consisting of Tus/Ter barrier based on the high-affinity binding of the E. coli protein, Tus, to specific DNA sequence called Ter; (2) Template Switching assay, focused on the capability of the proteins in reconstructing a functional LYS2 gene by TS. The obtained results demonstrated that (1) the absence of Rdh54 enhances replication fork regression, (2) Mte1 and Esc2 show opposite roles in telomere maintenance, (3) the interaction between Mte1 and Rad51 plays a crucial role in ensuring telomere stability and nuclear foci formation, (4) Mph1 and Mte1 promote cell survival through the break-induced replication (BIR) pathway. Further studies should assess the plausible interaction between Mph1 and Rdh54 proteins and characterize the function and interplay of the proteins involved in TS.

Animal coloration is as striking as it is diverse; however, the transcriptional basis of coloration is not deeply understood. Cichlid fishes are a tractable system for studying coloration as they exhibit a wide range of phenotypic diversity while remaining genetically similar. This facilitates the study of genotype-phenotype correlations and the identification of causative genes. RNA sequencing is a powerful approach to investigate the genes which characterize chromatophores. However, RNA-seq results can be plagued by the high abundance of rRNA in cells. This thesis aims to investigate differential gene expression between differently pigmented regions as well as explore the effects of tissue treatments and rRNA depletion on gene expression. Gene sets acquired with polyA selection, riboPOOL probes optimized for zebrafish, and zebrafish probes complemented with newly designed riboPOOL cichlid probes were compared to assess the functionality of these different rRNA depletion strategies. The use of zebrafish probes complemented with newly designed cichlid probes captured the greatest diversity of genes, many transcripts of which were missing from the other gene sets. Furthermore, as experiments such as scRNA-seq rely on a dissociation step, the effect of dissociation on gene expression was examined and found to promote the expression of stress response genes. The results of this upstream optimization were applied in the analysis of differential gene expression between the vertical stripes of the cichlid Pseudotropheus demasoni to better understand the molecular basis of vertical striping in fish. The dark stripes exhibited upregulation of melanic marker genes and the light, iridescent stripes showed an increase in iridophore marker gene expression. These findings were corroborated with cell count data from FACS to link transcriptional profiles and cell type quantifications. Overall, the study provides insight into the transcriptional basis of coloration in cichlid fishes and underscores the importance of optimizing methods drawing meaningful conclusions.

Human induced pluripotent stem cells (iPSCs) are an important in vitro model of disease and development. iPSCs can be differentiated in culture into cell types which are difficult to access from patients, such as neurons. Applying iPSC-derived cellular models to disease studies requires a thorough characterization of the derived cell types, as well as assessing reproducibility across cell lines or differentiation batches. With the aim of providing such a comprehensive molecular characterization at an early stage of cortical neuronal differentiation in vitro, six iPSC lines from four donors were differentiated to cortical neural progenitors using a modification of an established protocol (Shi et al., 2012a). The protocol successfully produced neural progenitors, with over 75% of the differentiated cells aligning with a cortical identity, as confirmed via qPCR and immunocytochemistry of established markers such as PAX6, NES and SOX1. To further classify the cell types produced as well as identify potential differences between cell lines, gene expression of the obtained cells was profiled with single cell RNA sequencing of ~22,000 cells, which uncovered the heterogeneity of neural progenitors produced. Further, although two differentiation batches produced similar cell-type compositions on a whole, a fraction of the lines showed inter-individual differences in cell type composition, which correlated with expression variability of known marker genes. Additionally, the cell types produced in vitro were compared to those produced in vivo by mapping our dataset to a reference fetal brain dataset (Polioudakis et al., 2019). It was observed that the in vitro dataset represented a subset of the cell types present at mid-gestation. Overall, the single cell characterization of differentiated cells allowed greater resolution in understanding cell-type heterogeneity of cortical neurogenesis, which is of key relevance for future applications such as disease modeling.

Acute leukemia is a life-threatening disease of blood and bone marrow, which is caused by malignant transformation of immature white blood cells. These malignant white blood cells invade space in bone marrow decreasing its ability to produce normal blood cells, eventually leading to death within weeks after the diagnosis without treatment. The acute leukemia can be broadly divided into its lymphoblastic and myeloid form, based on the affected cell lineage. Furthermore, acute leukemias can be classified based on different genomic features, such as gene fusions. Fusion genes are strong drivers in various cancers such as acute leukemias, and they are formed when two or more original genes join together forming a novel hybrid gene. If the novel hybrid gene is transcribed, it can lead to a translation of an abnormal fusion protein with altered function. The detection of the gene fusions is very important, since it affects to diagnosis and treatment of the patient. Various techniques can be used for fusion gene detection, of which the RNA sequencing is the method of choice, due to its ability to provide an unbiased identification of all known and novel gene fusions from the sample in a single experiment. In this thesis, the overarching aim was to develop an optimal sampling protocol for fusion gene detection using RNA sequencing for acute leukemia diagnostics. First, the whole blood samples in EDTA-tubes were collected from acute leukemia patients based on the findings from routine diagnostics. Next, the RNA was extracted at three different timepoints (0h, 8h, and 32h). The samples were stored at 4°C between the extractions. Finally, the RNA sequencing libraries were constructed, and the RNA sequencing was performed. After the sequencing, the data was analyzed using the FusionCatcher algorithm for fusion gene detection and the EdgeR-package for differential expression analysis. The FusionCatcher detected the same gene fusion in all the four fusion gene positive patients compared to routine diagnostics. However, the FusionCatcher failed to recognize the gene fusion in some of the samples with very low number of fusion breakpoint-spanning reads. These reads were visualized with IGV, suggesting that the detection failure resulted from the very low number of break-point-spanning reads. Furthermore, the sample storage did not affect on gene fusion detection. In addition, FusionCatcher detected PIK3AP::BLNK gene fusion from one of the fusion gene negative patients, suggesting a possibility that the patient truly was fusion gene positive. The differential expression analysis revealed changes in gene expression between the different timepoints. The results showed changes in various pathways related for example to cell death and protein biosynthesis, but also to pathways related to cancer. The results showed that prolonged sample storage alters the gene expression profile thus affecting the results of a gene expression study.

High-throughput sequencing techniques make it possible to identify DNA variants at a reasonable cost, representing a first-tier diagnostic test for rare mendelian diseases. However, a substantial number of variants identified through the analysis of sequencing data are frequently classified as variants of uncertain significance (VUS). Accordingly, only 30–60% of individuals receive a conclusive molecular diagnosis depending on the clinical phenotype. Reanalysis of older sequencing data has been encouraged by recently developed and improved methodologies for analysis and more robust bioinformatic pipelines to enhance variant interpretation and raise the diagnostic/detection rate. This study focused on reanalyzing data from a targeted gene panel, MYOcap, a targeted gene panel for patients with neuromuscular disorders. The aims were to find elusive (i.e., previously undetected/misinterpreted) variants in patients still missing a molecular diagnosis and, by using novel bioinformatic tools, focusing on pathogenic and likely pathogenic variants (according to ACMG guidelines) in Varsome as well as on variants affecting the splicing as predicted by SpliceAI. With this setting, the detection rate of solved cases increased by 2,7% in the first cohort and 0,5% in the third. This study suggests that additional data, such as segregation data or transcriptomic and proteomic data are essential for reducing the number of VUS and increase the detection rate. Notably, this study represents an essential first step of a larger reanalysis project, aiming at providing a diagnosis to an increasing number of myopathy patients.

Triple-negative breast cancer (TNBC) accounts for 10-15% of all breast cancer cases and has the worst clinical outcome. Characterizing features of TNBC are high recurrence and mortality rates, and the absence of three commonly targetable breast cancer biomarkers estrogen receptor, progesterone receptor, and HER2, limiting the number of targetable therapy options. Cytotoxic CD8 positive T cells play a crucial role in the anticancer immune response and act as a major component of successful cancer immunotherapies. However, cancer cells can evade T cell-mediated killing by overexpressing programmed death-ligand 1 (PD-L1) resulting in T cell exhaustion and limited immune response via the interaction with programmed death protein 1 (PD-1). Systemic anti-PD-L1/PD-1 therapies aim to prevent this immunosuppressive mechanism, but they are burdened with potentially life-threatening autoimmunity-type adverse effects. Therefore, cancer cell-specific targets to downregulate PD-L1 could offer efficacious and less harmful ways to overcome PD_L1/PD-1 mediated immunosuppression. Serine protease hepsin is commonly overexpressed in many solid tumors where it is responsible for the activation of HGF/MET signaling pathway as well as degradation of desmosomes and hemidesmosomes leading to the loss of epithelial integrity, invasion, and metastasis. Earlier studies have linked hyperactive HGF/MET pathway to the upregulation of immune checkpoint molecule PD-L1. In this thesis, I show how pharmacological inhibition of hepsin leads to decreased MET activity and downregulation of PD-L1 in a panel of TNBC cell lines. My results demonstrate the potential of hepsin-mediated regulation of PD-L1 in tumor immunosuppression, and hint at the potential of hepsin as a therapeutic avenue towards safe and efficacious immunotherapy in the future. These results are part of a larger study addressing the role of hepsin as a regulator of PD-L1 breast cancer.

Colorectal cancer is one of the most common cancers in the world, and in 2020 it was the cause of nearly 1 million deaths. A major reason for the high death rate is treatment resistance; eventually, almost all colorectal cancer patients with metastatic disease stop responding to chemotherapy. The problem of treatment resistance is not specific to this type of cancer, but it is a widespread issue for all cancer treatments. Chemotherapy resistance is the sum of several cellular and non-cellular factors that together enable sustained cell growth despite the treatment. The non-cellular factors are related to the tumor microenvironment, whereas the cellular factors are related to changes in gene expression, which facilitate e.g. the repair of drug-induced damage or lead to changes in drug metabolism. Lately, researchers have turned their interest to translational control in chemotherapy resistance. This is because translational control plays a major role in stress adaptation. During cellular stress, global translation rates are reduced and those messenger RNAs that are most important for cell survival are translated efficiently. Moreover, translation is fine-tuned by transfer RNA (tRNA) modifications. These modifications are chemical groups that are added to the ribose and the nucleobase of the tRNA molecule, and they affect all aspects of tRNA function, ranging from the structure and stability of the molecule to reading frame maintenance and rate of translation. tRNA modifications are dynamic and change in response to the cellular state, thus facilitating adaptation by translational control. Given the major role of translational control and tRNA modifications in cellular stress responses, their role in the chemotherapy response and adaptation should be thoroughly investigated. The aim of this thesis was to study how chemotherapy affects translation and tRNA modifications in a colon adenocarcinoma model. The cell lines SW480 (from a primary colorectal tumor) and SW620 (from a metastasis) were treated with 5-fluorouracil, oxaliplatin, and leucovorin (FOLFOX), a common combination of chemotherapeutics used in colorectal cancer treatment. The cells were subjected to long-term cyclic treatment as well as 24 h pulse treatment. Chemotherapy resistant cell lines were established by increasing the concentration of FOLFOX for each round of treatment. The effect on translation was studied by polysome profiling, which revealed that FOLFOX treatment causes immediate translational stress, as evidenced by the “shoulders” in the polysomal fractions in the profiles of the pulse treated cells. We hypothesized that these shoulders represent halfmers, polyribosomes without the large subunit. No difference was observed between the long-term treated cells and controls, possibly indicating that the cells had adapted to FOLFOX. The resistant cells exhibited slightly reduced translational activity, which might be due to altered function of ribosomes following the exposure to 5-fluorouracil. Changes in tRNA modification levels were quantified by liquid chromatography mass spectrometry. Several anticodon loop modifications exhibited altered levels after the pulse treatment. In addition, 5-FUrd, a metabolite of 5-fluorouracil, was incorporated into the tRNA. The long-term treated or resistant cells exhibited no differences in the modification levels. In conclusion, this study provided insights on the immediate effects of FOLFOX treatment on translation. This constitutes the first step towards understanding how RNA-based regulatory mechanisms may contribute to the effect and possible resistance to chemotherapy.