Optimized Analytical Procedures for the Untargeted Metabolomic Profiling of Human Urine and Plasma by Combining Hydrophilic Interaction (HILIC) and Reverse-Phase Liquid Chromatography (RPLC)-Mass Spectrometry

Abstract

Profiling of body fluids is crucial for monitoring and discovering metabolic markers of health and disease and for providing insights into human physiology. Since human urine and plasma each contain an extreme diversity of metabolites, a single liquid chromatographic system when coupled to mass spectrometry (MS) is not sufficient to achieve reasonable metabolome coverage. Hydrophilic interaction liquid chromatography (HILIC) offers complementary information to reverse-phase liquid chromatography (RPLC) by retaining polar metabolites. With the objective of finding the optimal combined chromatographic solution to profile urine and plasma, we systematically investigated the performance of five HILIC columns with different chemistries operated at three different pH (acidic, neutral, basic) and five C18-silica RPLC columns. The zwitterionic column ZIC-HILIC operated at neutral pH provided optimal performance on a large set of hydrophilic metabolites. The RPLC columns Hypersil GOLD and Zorbax SB aq were proven to be best suited for the metabolic profiling of urine and plasma, respectively. Importantly, the optimized HILIC-MS method showed excellent intrabatch peak area reproducibility (CV < 12%) and good long-term interbatch (40 days) peak area reproducibility (CV < 22%) that were similar to those of RPLC-MS procedures. Finally, combining the optimal HILIC- and RPLC-MS approaches greatly expanded metabolome coverage with 44% and 108% new metabolic features detected compared with RPLC-MS alone for urine and plasma, respectively. The proposed combined LC-MS approaches improve the comprehensiveness of global metabolic profiling of body fluids and thus are valuable for monitoring and discovering metabolic changes associated with health and disease in clinical research studies.

Fig. 1.

Optimization of the HILIC-MS analytical procedure. (A) Individual scores of standards under the different HILIC conditions (related to Fig. S1). The best score in positive and negative ESI modes was selected. (B) Score of the metabolic features from a urine sample under the different HILIC conditions. Metabolic features from positive and negative ESI modes were combined (related to Fig. S2). (C) Repartition of the metabolic features (combination of positive and negative ESI modes) along the chromatographic runs (related to Fig. S2).

Fig. 2.

Suitability of the optimized HILIC-MS procedure for the untargeted analysis of hydrophilic metabolites in urine and plasma. Number of metabolites labeled as detected and quantified in HMDB in urine, blood, and both classified by chemical classes (left panel). The score of 137 standards belonging to the different chemical classes under the optimized HILIC-MS approach is shown on the right panel; 93% of the standards had a good or acceptable score.

Fig. 3.

Optimization of the RPLC-MS analytical procedure. (A) Individual scores of standards with increasing calculated logP values under the different RPLC conditions in negative mode. Score of the metabolic features from (B) urine and (D) plasma under the different RPLC conditions. Repartition of the metabolic features from (C) urine and (E) plasma along the chromatographic runs.

Fig. 4.

Intra- and long-term interbatch reproducibility of the optimized HILIC- and RPLC-MS procedures. (A) Intrabatch and long-term interbatch reproducibility of the retention time of metabolic features across two batches analyzed 42, 38 days and 71 days after an initial analysis with ZIC-HILIC, Hypersil GOLD and Zorbax SB aq columns, respectively. Each batch consisted of 10 injections of the same urine sample for ZIC-HILIC and Hypersil GOLD columns and of 10 injections of the same plasma sample for Zorbax SB aq column (related to Fig. S6). (C) Intra- (light color) and long-term interbatch (dark color) reproducibility of peak area (related to Fig. S7). The horizontal black line represents the CV median and the red diamond the CV mean.

Fig. 5.

Combination of HILIC- and RPLC-MS in positive and negative ESI modes and expansion of the metabolome coverage of human urine and plasma. (A) Venn diagrams representing the proportion and quantity of metabolic features detected only in HILIC mode compared with RPLC alone for urine (top panel) and plasma (bottom panel) samples in positive (left panel) and negative (right panel) ESI modes (related to Fig. S8). (B) Individual scores of the standards that had unacceptable scores or that eluted in the void volume in HILIC mode when injected onto Hypersil GOLD column (related to Fig. S9).