(23d) Investigating the Neurodevelopmental Exposome: A Proposed Transcriptomic and Metabolomic Analysis of Mother-Child Cohort Pairs in Portugal

The human exposome is comprised of incredibly complex and dynamic exposure scenarios that includes every exposure that an individual is subjected to during their entire lifespan. As such, the exposome, and therefore the effect it may exert, are in constant flux, which makes quantifying and predicting the effects very difficult. Regardless, understanding the exposome is instrumental to understanding the connection between the environment and human health and the mechanisms behind many diseases that can lead to death, illness, and overall decreases in quality of life. Omics techniques are frequently used to do this. Omics is the study of the complex interactions that can influence the phenotype of an organism. When combined, omics has been instrumental in a host of applications including disease discovery. The present study will utilize two omics technologies and combine them in a multi-omics approach. Transcriptomics is the study of the transcriptome, which is the complete set of RNA that is produced by the genome of an organism under specific conditions. In general, RNA are the templates for protein synthesis and therefore reflect the genes that are actively expressed. Microarrays are a frequently used technology in analyzing the RNA in an organism and can assess tens of thousands of genes in a single application, depending on the target RNA of interest. To do so, fluorescently labelled RNA are hybridized to an array with a set of complementary short nucleotide oligomers (“probes”). The resulting fluorescence intensity emitted from each probe location is indicative of the transcript abundance for that probe sequence. The genes associated with each probe can then be determined and information such as differentially expressed genes (DEGs) amongst conditions can be assessed. In an associated analysis (Schultz et al 2022), a statistical pipeline was generated to reliably uncover DEGs. Conversely, metabolomics is the study of the small molecules that result from the endogenous breakdown of food, drugs, chemicals, or bodily tissues. Metabolomics is the most recently developed omics technique and has been instrumental in bolstering the findings from other omics analyses. Further, because metabolites are the end product of gene expression, it is thought to be a more sensitive endpoint to measure biological phenotype. However, metabolomic changes can occur rapidly, which makes it difficult to distinguish constant versus transient profile signatures. Within the framework of the HEALS project, a multi-disciplinary approach has already been developed to investigate the link between prenatal exposure to endocrine disruptors and early developmental health outcomes in children. In house development of untargeted metabolomics workflows have previously been completed (Sarigiannis et al 2021; Papaioannou 2018). This pipeline has proven effective in generating a reproducible and reliable workflow while being able to deal with large numbers of samples and massive amounts of generated data. However, while the metabolomic analysis pipeline is well established, there is still the need to develop a reliable transcriptomic workflow and integrate the resulting data. In the present proposed study and as a part of the HEALS European Exposure and Health Examination Survey (EXHES), 548 pregnant mothers and their children were recruited. Biosamples and questionnaire data was collected. The aim of the study is to link the data generated from both metabolomic and transcriptomic datasets and questionnaire data using multi omics and correlation analysis.

For metabolomic analysis, serum samples were stored in screwtop cryovials and placed at -80℃ until later analysis. Samples were thawed in stable conditions overnight on the day before analysis at 4℃. Each sample was briefly vortexed to homogenize (5 s at 2000 rpm), 200 μl was transferred to a new 1.5 mL Eppendorf tube, and 50 μL was used to prepare a pooled quality control (QC) samples which followed the sample extraction protocol. After aliquoting of all samples, 600 μL of pre-chilled methanol was added to each sample for protein precipitation and metabolite extraction (1:3 sample to methanol ratio). Samples were stored at -20℃ for 30 minutes to facilitate protein precipitation and then centrifuged (15000 rpm for 15 min at 4 ℃). 300 μl of the supernatant was transferred to a new 1.5 mL Eppendorf tube and concentrated by nitrogen blown down evaporation at a gentle flow. Dried extracts were reconstituted LC-MS grade water and vortexed (5 min at 2000 rpm). QC samples were reconstituted with LC-MS grade water with 1 ppm internal standard additions. Samples were centrifuged (15000 rpm for 10 min at 4 ℃) and 100 μL of the supernatant was transferred to autosampler vials. Sample randomization was performed to control for batch effects.

Untargeted metabolomic analysis of serum samples will be performed using Agilent 1290 Infinity HPLC LC System coupled to an Agilent 6540 HRMS-QTOF/ LCMS system. Mobile Phase A was LC-MS grade water and mobile phase B was methanol, both with 0.1% formic acid, for positive and negative ionization mode. Reverse phase HPLC chromatography technique was applied using a Fortis SpeedCore pH+ C18 column (2.1 x 100 mm, 2.6 μm, Fortis Technologies, United Kingdom). The flow rate was set at 0.250 mL/min at a steady temperature of 40℃. The following gradient elution program was the same for both positive and negative mode: 0% B at 2 min, 100% B at 17 min, 100% B at 22 min, 0% B at 24 min, and 0% B at 26 min. An injection volume was 5 μL used. The data are acquired between 50 and 1000 m/z at a scan rate of 1.4 spectra/s in centroid mode at a resolution of 40,000 FWHM. The source conditions of the Q-TOF system are as follows: gas temperature 300℃, drying gas 7 L/min, nebulizer 50 psig, fragmentor 150 V, skimmer 65 V, and capillary voltage 3500V or -3500V in positive or negative mode, respectively.

The workflow for metabolomic data analysis followed the established in-house pipeline. The data from MS/MS analysis is processed using the Bioconductor R - based packages XCMS and CAMERA. Initially, the raw data files (.d format) are converted to .mzML using the MSConvert GUI tool included in the ProteoWizard v.3.0.20270 software. The data is then imported into RStudio and the base peak chromatograms (BPC) across all samples is inspected. Boxplots are created to display the distribution of total ion currents (TIC) per file, and heatmaps are extracted and correlated based on the similarity of their BPC. Before running the chromatographic peak detection using the centWave algorithm, the optimization of xcms parameters is required. By adding internal standards to the QC samples, the two most critical parameters (peakwidth and expected mass error (ppm)) are evaluated. The Obiwarp method is used to align the samples and evaluate the alignment by plotting the differences of the adjusted- to the raw- retention times per sample. To complete the spectral pre-processing, correspondence is completed using the peak density and the featureDefinitions methods. The fillChromPeaks method is used to fill in intensity data for missing values from the original data. Then, a batch effect correction is applied to reduce variability across the data. An 80% rule is applied to the QC samples and the median RSD is calculated for all the detected metabolites. This determines the instrument and overall process variability. Network-based annotation is performed using the xMSannotator in package that includes the multilevel annotation function to perform metabolite annotation, which associates metabolites detected by MS/MS to known chemicals and denotes them with different confidence levels. This function uses online databases such as HMDB, KEGG and LipidMaps. The annotation is completed for positive and negative ionization modes separately. The optimized calculated mass errors are set as mass tolerance in ppm for database matching. The adduct list used for database searching is set at c("M+H") and c("M-H") for positive and negative adducts, respectively. Only for the HMDB database is the metabolites status set at "Detected", and the biofluid location turned to "blood". Integrated pathway analysis is achieved using Mass Profiler Professional v.14.9 software. The metabolic pathways are determined by searching through the MetaCyc, Wikipathways, and KEGG databases using Fisher's exact test. The Exposome-Wide Association Study (EWAS) approach was initially implemented to systematically explore and associate multiple exposure factors and modifiers, and so to discover and replicate robust correlations with metabolite levels and dysregulated pathways. The 'X-Wide Association Analyses (XWAS)' package in R will be used to achieve these outcomes. But, while the metabolomic analysis pipeline is well-developed, the challenge of the current proposed research is to develop and integrate the analysis of the transcriptomic data into the multi-omic workflow.