Организация, в которой проходила защита:
Институт биоинформатики (Санкт-Петербург)
Год защиты:2018
Аннотация:The Russian Children’s Study is a prospective cohort of 516 boys who were enrolled at 8–9 years of age and provided semen samples at 18–19 years of age. RRBS of sperm was conducted to identify the methylation level of CpG denucleotides. At the moment of enrollment into the study, the TCDD dioxine (which is one of the most harmful endocrine disrupting chemicals) concentration in the blood of each boy was measured for further longitudinal study of its influence on the reproductive health. Moreover, each boy visited the clinic biennially - for blood sampling; annually - for urine sampling, follow up of growth and puberty and interviewing => 20 000+ sample aliquots and 1000+ analyzing parameters were collected in total for further analysis.
What is already known?
• Peripubertal exposure to TCDD is associated with poorer semen quality (RCS, Minguez-Alarcon et al., 2017)
• 52 differentially methylated regions (DMRs) were identified that distinguished lowest and highest peripubertal serum TCDD concentrations (RCS, Pilsner et al., 2018)
What is needed to be known?
• How do other factors influence the methylation level of the human sperm?
• Which factors can mediate the effect of the peripubertal exposure to TCDD?
Aim of the project:
Mediation analysis using regression models and longitudinal design
Predictors: peripubertal TCDD concentration and smoking
Outcomes: sperm methylation profiles (all CpGs with coverage >=10 using RRBS)
What were the data to analyse:
• 34 samples with different concentrations of TCDD in the boy’s serum at enrollment in the study (8-9 years old) (you can see the histogram of TCDD concentration destribution on the figure 1).
• Methylation level of 2 611 773 CpGs with coverage >=10 presented in at least one of 34 samples
• Data regarding lifestyle habits of each of 34 chosen participants (you can see the list of questions based on which we evaluated smoking within 6 months before sperm collection as a range variable with 6 categories on the figure 2 and bar chart with the count of partisipants belonging to each category on the figure 3).
What was going to be done with these data:
Building the linear regression model of TCDD concentration and methylation level
Regress the smoking on the TCDD (predictor) to recognize whether the TCDD is a significant predictor of the mediator. If the mediator is not associated with the TCDD level, then it could not possibly mediate anything and then these two variables can be used only as independent predictors in multi-factorial regression model.
Regress the CpGs methylation on smoking to confirm that it is a significant predictor of the CpG methylation.
Building multi-factorial regression model of TCDD concentration in the boy’s serum at enrollment in the study and smoking within 6 months before sperm collection either as two independent predictors or as a predictor and mediator (based on the result of n.2) and methylation level of sperm as a dependent variable.
Collect CpGs whose methylation level is significantly dependent on TCDD concentration and/or smoking (three models in total).
Mapping of all significant CpGs found in each model to the human genome
Biological sense analysis of significant CpGs found (functional enrichment analysis)
Brief summary of the results:
Linear regression models for the influences of TCDD concentration in prepubertal age and smoking within 6 months before collection of sperm on the young adults semen methylation level were built (both for separate predictors and for their combination).
Predictors were considered to be independent on each other.
Only CpGs whose methylation levels were significantly dependent on the TCDD concentration and/or smoking, were chosen for further analysis.
All significant CpGs were mapped to either gene, promoter or enhancer regions.
Enrichment analysis for all genes associated with CpGs, whose methylation levels were significantly dependent on one or both of predictors, was performed.
Some interesting findings are needed to be analyzed in more details.