(160ae) Development of an Advanced Plasma Untargeted Metabolomics Method for the Characterization of the Human Exposome.

Mapping the human exposome becomes a challenge as it is constantly changing and varies according to the stage of an individual's life. However, exposome studies could make us understand the connection between environmental exposure and human health and help us discover the causes of various diseases by exploring the interactions between genetic predisposition and exposure, both external and internal. In the framework of the EU project HEALS we developed a multi-disciplinary and holistic approach to access prenatal exposure to endocrine disruptors and gain insight into their early childhood development effects. Our state-of-the-art untargeted metabolomics workflow, including data acquisition, data pre-processing, and data processing, aiming to overcome the obstacle of the lack of standardized protocols, as well as the challenges resulting from the analysis of a large number of human samples (>100). The analysis was applied on 173 plasma samples of pregnant women at delivery collected in the HEALS pilot European Exposure and Health Examination Survey (EXHES).

Plasma samples were stored in Eppendorf tubes in the freezer at -80 ° C. One day before analysis, they were thawed and stored in stable conditions at 4 ℃. Firstly, the samples were vortexed to homogenize for 5s at 2000 rpm, and 200 μl of each sample was placed in the new Eppendorf. Furthermore, a quantity of 50 μL of each sample was used to prepare the pooled quality control (QC) sample. 600 μL of 1:3 pre-chilled methanol was added to each sample for protein removal and metabolite extraction. After centrifugation, at 15000 rpm for 15 min at 4 ℃, 300 μl of the supernatant was transferred to a new Eppendorf, following by evaporation to dryness through a sample concentrator under a gentle flow of N₂. The dry extracts were reconstituted with 100% water LC-MS grade with internal standards and vortexed for 5 min at 2000 rpm. Finally, the samples were centrifuged at 15000 rpm for 10 min at 4 ℃, and 100 μL of the supernatant was transferred to autosampler vials. The same procedure was followed for the preparation of the pooled QC samples.

The untargeted metabolomic analysis of plasma samples was 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 with 0.1% formic acid, and mobile phase B was methanol with 0.1% formic acid, both for positive and negative ionization mode. The chromatography technique applied was reverse phase HPLC, 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 0.350 mL/min at a constant temperature of 40 ^oC. 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. The injection volume was 5 μL. 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 the following: gas temperature 300 ^oC, 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 data from MS/MS analysis was processed using the Bioconductor R - based packages XCMS and CAMERA. Firstly, the raw data in .d format was converted to .mzML format using the MSConvert GUI tool included in the ProteoWizard v.3.0.20270 software, followed by the data import. An initial data inspection was performed, including the base peak chromatograms of all samples, as our data's first evaluation. Furthermore, boxplots were created displaying the distribution of total ion currents per file, and heatmaps were extracted, grouping the samples based on the similarity of their base peak chromatogram. Before running the chromatographic peak detection using the centWave algorithm, the optimization of xcms parameters was required. By adding internal standards to the QC samples, the two most critical parameters, peakwidth and expected mass error (ppm), were evaluated. We used the Obiwarp method to align the samples and evaluated the alignment by plotting the differences of the adjusted- to the raw retention times per sample. The final step in the spectral pre-processing was correspondence using the peak density and the featureDefinitions methods. The fillChromPeaks method was used to fill in intensity data for missing values from the original data. Then, a batch effect correction was performed, to reduce variability in our data. We applied the 80% rule to the QC samples and calculated the median RSD for all the detected metabolites. Thus, the instrument and overall process variability were determined. The Network-based annotation was performed using an R package, xMSannotator. This package includes the multilevel annotation function to perform metabolite annotation. The purpose of this function is to associate metabolites detected by MS/MS to known chemicals into different confidence levels, using online databases such as HMDB, KEGG and LipidMaps. The annotation was achieved both on positive and negative ionization modes. The optimized calculated mass errors were set as mass tolerance in ppm for database matching. We set the adduct list used for database searching at c("M+H") and c("M-H") for positive and negative adducts, respectively. Only for the HMDB, the metabolites' status was set at "Detected", and the biofluid location was turned to "blood". Integrated pathway analysis was achieved using Mass Profiler Professional v.14.9 software. The metabolic pathways were determined by searching through the MetaCyc, Wikipathways, and KEGG databases using Fisher's exact test. Also, we performed a quantitative analysis of T manganese, iron, cadmium, strontium, selenium, mercury and zinc. The Exposome-Wide Association Study (EWAS) approach was adopted to comprehensively explore and associate multiple exposure factors and modifiers, discovering and replicating robust correlations with metabolites levels and dysregulated pathways. The 'X-Wide Association Analyses (XWAS)' package in R was used for this. Volcano plots and correlation globes were used for data visualization.

Out of the 2415 detected metabolites, the metabolite identification results pointed out that the total number of annotated metabolites in the plasma samples was 1274. The total number of the pathways in which the detected metabolites are involved was 190. The association of the metabolites with the studied metal concentrations revealed the correlation of plasma biomarkers such as citric acid, L-tyrosine, and L-phenylalaline, with human exposure to metals. The perturbated values of citric acid led to the disruption of the TCA cycle. At the same time, L-phenylalaline, L-tyrosine, L-tryptophan, and L-glutamine affected the amino acid metabolism pathway.

In conclusion, the present metabolomic study suggests the disruption of central biochemical processes; thus, the dysregulation of the mechanism of the maintenance stability of cells' internal energy after exposure to metals could cause adverse outcomes to infant development. This information, combined with other omics technologies, could contribute to developing new methods for the prevention and the improvement of risk exposure.