Integrated Methylation/isomiR/gene Expression (MIG) bio-profiles for prediction of treatment response in rheumatoid arthritis
We are developing tools to analyze next generation sequencing data collected from rheumatoid arthritis (RA) patients in Norway. In 2017 we continued to investigate variation in miRNAs and the impact on regulation. We have also been looking at how this varies between human and mice, to understand the relevance of RA mice studiesOur goal is to investigate changes in features that are responsible for regulating gene, and hence protein, expression. We are trying to combine three different kinds of data: miRNAs levels and methylation events, which influence gene expression levels; and mRNA levels that are a measure of gene expression. So, if we see that miRNA A is more highly expressed in a group of RA patients, and we know that this miRNA down regulates gene B, then we would expect to see lower levels of the mRNA corresponding to gene B. However, the major challenge we are facing in this project is there is much lower patient enrolment than anticipated in the partner project that seeks to identify relevant RA patients, isolate samples and collect the data to give us miRNA and methylation profiles. This lower number of patients means is that is it more difficult to differentiate real differences in miRNA and mRNA expression levels from random fluctuations. In general, the larger the number of patients in a cohort, the more information and better statistical support to help make this distinction. However, one of the advantages of our project is that our focus is the development of novel methods to investigate these combined patient profiles (i.e., miRNA levels, methylation events, and mRNA levels). So, although the cohort we ultimately plan to study is small and patients are still being enrolled, we can develop our methodology using more comprehensive publicly available datasets and understand how different factors contribute to the variation. This can originate from various steps in the experimental process, or even from decisions that are made in the subsequent analysis. Thus, we have been investigating a set of more than 400 samples from both Human and Mouse studies, in particular from the The Cancer Genome Atlas (cancergenome.nih.gov) as many of these data are typically collected by a single lab and this removes one source of variability. Last year, we reported that, in addition to looking at variation in miRNA levels, we were also investigating isomiRs, which are variations in the miRNA sequence (either extensions, deletions, point mutations or a combination of these possibilities) and developing standards for quantifying these populations. These variations are important because an isomiR can regulate a target gene that is different from its "parent" miRNA. Thus, it is important to also look for variations in isomiR populations between two conditions. As a side project we are investigating how these populations vary between human and mouse. This is also important because mice are often used in RA studies and differences in miRNA and their isomiR populations can significantly impact the relevance of results obtained in mice studies. On the other hand, if we find that a particular miRNA has the same isomiR population (or profile) in both human and mouse, and we additionally find (via a mouse study) that this miRNA regulates a gene that is important in RA, then it indicates that this study and result could be relevant to human. We are in the process of finishing the study and preparing a manuscript for submission. We then plan to use our findings in the analysis of the RA data from the partner project.
We are developing analytical tools to investigate next generation sequencing (NGS) data collected from rheumatoid arthritis patients in Norway. We have created a tool for investigating variation in short RNAs, which are responsible for gene regulation and tested it against these NGS data. This is the first major milestone for our project.Rheumatoid arthritis (RA) is a chronic inflammatory disorder of the joints occurring in ~1% of the Norwegian population. Treatments are available but commonly associated with major side effects and effectiveness varies among subjects. Genetic disposition and environmental factors are thought to be associated with both the onset of RA and response to treatment and we are trying to establish and clarify these roles. In this way it may be possible to predict severity of onset and optimal treatment for a patient. Our project is an extension of existing HSØ funded RA studies involving Benedicte Lie, also in the department of medical genetics at OUS, and Diakonhjemmet and Martina Hanssen Hospitals who are enrolling new patients to the study. Full details of the patient enrollment and study design are provided in the eReport “MicroRNAs as biomarkers for treatment response in rheumatoid arthritis”. In brief, there are two study groups comprising (i) newly diagnosed patients, and at three months treatment with methotrexate and (ii) patients receiving methotrexate treatment who have been in remission for more than a year. In Lie’s project, they are investigating variation in microRNAs (miRNAs) levels. MiRNAs are short (~22nt) RNAs that influence gene regulation; a change in the level of a miRNA can effect a change in downstream expression of its target genes. By looking for significant changes in miRNA levels between conditions and examining/predicting their targets it may be possible to identify associations with disease state. However, the reality is more complex - many miRNAs actually exist as a population of modified forms, or isomiRs, manifested as additions or deletions at either end of the sequence, or as mutations within the sequence. A major milestone is to characterize these isomiR populations, understand how they vary among conditions and interpret the significance of these changes. For example, a major change in isomiR population may impact the target genes that will be regulated. As a first step we have defined a system to define the types of isomiRs that can be present in a miRNA population. From this, we can characterize a miRNA, in terms of its isomiRs, and investigate how the population varies among conditions. In addition, we have been developing the necessary software tools for analysing small RNA NGS datasets to extract isomiR data, identify significant isomiR population changes (IPCs) amongst conditions, and estimate the impact of these changes. A preliminary set of RA NGS data is available from Lie’s project, and we have also been working with other publicly available data to test the software and learn what kind of IPCs we might expect. We have also been investigating changes in IPCs associated with different experimental protocols. This is crucial, as we need to distinguish biological IPCs from experimental artefacts. Our initial results are interesting. We have identified significant changes in isomiR populations between cancer and healthy subjects in two different datasets. Specifically, we consistently found that cancer subjects appear to have much narrower isomiR populations compared to healthy subjects. The likely impact of these differences is that cancer patients will have a much less noisy gene regulation network. As more RA small RNA data is collected, we hope to have sufficient power to resolve similar differences in patient data. This will form the primary focus of our work in 2017.