LC/MS-based discrimination between plasma and urine metabolomic changes following exposure to ultraviolet radiation by using data modelling

Abstract

Introduction: This study sought to compare between metabolomic changes of human urine and plasma to investigate which one can be used as best tool to identify metabolomic profiling and novel biomarkers associated to the potential effects of ultraviolet (UV) radiation.

Method: A pilot study of metabolomic patterns of human plasma and urine samples from four adult healthy individuals at before (S1) and after (S2) exposure (UV) and non-exposure (UC) were carried out by using liquid chromatography-mass spectrometry (LC-MS).

Results: The best results which were obtained by normalizing the metabolites to their mean output underwent to principal components analysis (PCA) and Orthogonal Partial least squares-discriminant analysis (OPLS-DA) to separate pre-from post-of exposure and non-exposure of UV. This separation by data modeling was clear in urine samples unlike plasma samples. In addition to overview of the scores plots, the variance predicted-Q2 (Cum), variance explained-R2X (Cum) and p-value of the cross-validated ANOVA score of PCA and OPLS-DA models indicated to this clear separation. Q2 (Cum) and R2X (Cum) values of PCA model for urine samples were 0.908 and 0.982, respectively, and OPLS-DA model values were 1.0 and 0.914, respectively. While these values in plasma samples were Q2 = 0.429 and R2X = 0.660 for PCA model and Q2 = 0.983 and R2X = 0.944 for OPLS-DA model. LC-MS metabolomic analysis showed the changes in numerous metabolic pathways including: amino acid, lipids, peptides, xenobiotics biodegradation, carbohydrates, nucleotides, Co-factors and vitamins which may contribute to the evaluation of the effects associated with UV sunlight exposure.

Conclusions: The results of pilot study indicate that pre and post-exposure UV metabolomics screening of urine samples may be the best tool than plasma samples and a potential approach to predict the metabolomic changes due to UV exposure. Additional future work may shed light on the application of available metabolomic approaches to explore potential predictive markers to determine the impacts of UV sunlight.

Keywords: LC–MS; Metabolomic profiling; Orthogonal partial least squares- discriminant analysis (OPLS-DA); Plasma; Principal components analysis (PCA); Ultraviolet radiation (UV); Urine.

Fig. 1

Indicative representation of plasma and urine collection schematic at two consecutive conditions, control conditions, without a dose of UVA light, (UC) and Ultraviolet-A conditions, with a dose of UVA light, (UVA). The first urine sample on each day was the first pass (typically 6 am–8 am)

Fig. 2

Score Plot for first (S1) and second (S2) samples for each of non-exposure (UC) and exposure-ultraviolet (UV) groups by using PCA (R2X(cum) 0.660, Q2(cum) 0.429, two components) based on the identified metabolites from positive and negative ion modes in Plasma samples

Fig. 3

Overview of the observations of OPLS-DA model for separation between two non-exposure samples (S1UC and S2UC) in plasma from healthy subject. R2X(cum) 0.959, R2Y(cum) 1.0, Q2(cum) 0.968

Fig. 4

Overview of the observations of OPLS-DA for plasma metabolites with positive and negative ion modes to show the separation between pre- and post-UV exposure samples (S1UV and S2UV). R2X(cum) 0.944, R2Y(cum) 1.0, Q2(cum) 0.983

Fig. 5

Cross validation of OPLSDA model for the classification of non-exposure group (UCS1-UCS2) in plasma samples by the Permutations test

Fig. 6

Cross validation of OPLSDA model for the classification of UV-exposure group (UVS1-UVCS2) in plasma samples by the Permutations test

Fig. 7

Score Plot for first (S1) and second (S2) samples for each of non-exposure (UC) and exposure-ultraviolet (UV) groups by using PCA (R2X(cum) 0.982, Q2(cum) 0.908, seven components) based on the identified metabolites from positive and negative ion modes in urine samples

Fig. 8

Overview of the observations of OPLS-DA model for separation between two non-exposure samples (S1UC and S2UC) in urine from healthy subject. R2X(cum) 0.892, R2Y(cum) 0.998, Q2(cum) 0.996

Fig. 9

Cross validation of OPLSDA model for the classification of non-exposure group (UCS1-UCS2) in urine samples by the Permutations test

Fig. 10

Overview of the observations of OPLS-DA for urine metabolites with positive and negative ion modes to show the separation between pre- and post-UV exposure samples (S1UV and S2UV). R2X(cum) 0.914, R2Y(cum) 1.0, Q2(cum) 1.0

Fig. 11

Cross validation of OPLSDA model for the classification of UV-exposure group (UVS1-UVS2) in urine samples by the Permutations test